Apparatus and method for motion compensation

The present invention relates to manipulating video streams.

The Discrete Cosine Transform

A two dimensional DCT converts an input matrix of N×N spatial domain elements to a matrix of N×N DCT coefficients. In many standard video compression schemes N=8. A two dimensional DCT may be implemented by applying a one dimensional DCT over the rows of the input matrix to provide a row-transformed matrix and them applying a one dimensional DCT over the columns of the row-transformed matrix. Those of skill in the art will appreciate that the same result is achieved when applying a one dimensional DCT over the columns of the input matrix to provide a column-transformed matrix and then applying a one-dimensional DCT over the column-transformed matrix to provide the matrix of N×N DCT coefficients.

A plurality of fast algorithms for performing the DCT was introduced by C. Loeffer, A. Ligtenberg and G. S. Moschytz (“Practical fast 1−d DCT algorithms with 11 multiplications”, in Proceedings of ICASSP 1989, pp. 988-991.). Loeffer et al suggest a four stage DCT for conversion of input matrix of 8×8. Each stage is executed in series while the calculation within each stage may be calculated in parallel. The mentioned above article includes a graphical illustration of four-stage algorithms. One of their algorithms may also be illustrated by the following sets of equations:

A 0= I 0+ I 7 ; A 1= I 1+ I 6 ; A 2= I 2+ I 5 ; A 3= I 3+ I 4 ; A 4= I 3− I 4; A 5= I 2 −I 5 ; A 6= I 1− I 6; A 7 =I 0 −I 7.

B 0= A 0+ A 3; B 1= A 1+ A 2; B 2= A 1= A 2; B 3= A 0− B 3; B 4, B 7=ROT( C 3)[ A 4, A 7]; B 5, B 6=ROT( C 3)[ A 5, A 7];

C 0= B 0+ B 1; C 1= B 0− B 1; C 2, C 3=ROT(√{square root over (2)} C 1)[ B 2, B 3]; C 4= B 4+ B 6; C 5= B 7− B 5; C 6= B 4− B 6; C 7= B 5+ B 7;

O0=C0; O 1= C 7+ C 4; O2=C2; O 3=√2 C 5; O4=C1; O 5=√2 C 6; O6=C3; O 7= C 7 −C 4

Whereas each of these equation sets corresponds to a single stage, I 0 –I 7 are the inputs signals to the DCT transform, O 0 –O 7 are further divided by the constant √{square root over (8)} to provide the outputs of the DCT, A 0 –A 7 are the intermediate results of the first stage of the DCT; B 0 –B 7 are the intermediate results of the second stage of the DCT; C 0 –C 7 are the intermediate results of the third stage of the DCT; and the ROT denoted a rotation operation. A rotation operation by k C n is illustrated by the following equations, whereas E 1 and E 2 are the inputs of the rotation operation while F 1 and F 2 are the outputs of the rotation operation:

F 1= E 1* k *cos( nπ/ 2 N )+ E 2 *k * sin( nπ/ 2 N )

F 2=− E 1 *k *sin( nπ/ 2 N )+ E 2 *k *cos( nπ/ 2 N )

The upper equation is also referred to as UP_ROT, whereas the lower equation is also referred to as LOW_ROT.

Loeffer et al suggested four other first stages that are illustrated by the

following equation sets:

A 0 =I 0 +I 7 ; A 1= I 1+I 2 ; A 2= I 1− I 2; A 3= I 3+ I 4 ; A 4= I 3 I 4 ; A 5= I 5+ I 6 ; A 5= I 5 −I 6 ; A 7 =I 0 −I 7.

A 0= I 0 +I 7 ; A 1= I 5+ I 1 ; A 2 =I 6+ I 2 ; A 3= I 3+ I 4 ; A 4= I 3 −I 4 ; A 5= I 1 −I 5 ; A 6= I 2 −I 6 ; A 7 =I 0 −I 7.

A 0 =I 0 +I 3 ; A 1 =I 6+ I 1; A 2 =I 5 +I 2 ; A 3 =I 0 −I 3 ; A 4 =I 7 +I 4 ; A 5 =I 2 −I 5 ; A 6 =I 1 −I 6 ; A 7 =I 4 −I 7.

A 0= I 0 +I 4 ; A 1 =I 6 +I 1; A 2 =I 5 +I 2 ; A 3 =I 3 +I 7 ; A 4 =I 0 −I 4 ; A 5 =I 2 −I 5 ; A 6 =I 1 −I 6 ; A 7 =I 3 −I 7.

For convenience of explanation these first stages are referred to as S 12 , S 13 , S 14 and S 15 respectively.

Loeffer et al further suggests to reverse the second and third stages of the even part (the part that calculates outputs O 1 , O 3 , O 5 and O 7 ). They also suggested sixteen additional combinations of a sequence that includes the second, third and fourth stages of the even part.

It is noted that the inverse DCT (IDCT) can be implemented by the same algorithms, but in reverse order.

Video Compression Schemes

Digital video must be extensively compressed prior to transmission and storage, as each picture includes multiple pixels, and each pixel has three color-difference multi-bit values.

Standard compression schemes (such as the MPEG compression standards, JPEG, H.263 and others) utilize multiple compression techniques to achieve a very significant compression ratio.

JPEG compression scheme includes the following steps: (i) color space conversion—converting a matrix of RGB pixel values to a matrix of luminance and chrominance values (YUV); (ii) Spatial transform of applying a Discrete Cosine Transform (DCT) upon the YUV matrix to provide a matrix of frequency coefficients, each frequency coefficient describes how much of a given spatial frequency is present; (iii) quantization, during which each spatial coefficient is divided by a quantizing factor such that small spatial coefficients are truncated to zero; (iv) zig-zag scanning and run-length coding the quantized matrix, for achieving a compressed representation of the quantized matrix, as a typical quantized matrix includes many zero-value coefficients; and (v) variable length coding, such as Huffman coding, to provide a compressed matrix.

K. Froitzheim and H. Wolf, “A knowledge based approach to JPEG acceleration”, IS&T/SPIE Symposium on Electronic Imaging: Science and Technology. San Jose, USA, Feb. '95. Offers an algorithm to reduce the complexity of JPEG compliant image processing. Froitzheim et el suggests to reduce the amount of operations involved in calculating IDCTs by applying the IDCT transform only to rows or columns of a matrics of DCT coefficient (or semi-transformed matrix) that include at least one non-zero element. Accordingly, before applying a row-wise IDCT, the rows of the matrix are checked to locate rows that include at least one non-zero element. Before applying a column-wise IDCT, the columns of the matrix are checked to locate columns that include at least one non-zero element.

Various compression schemes (such as the MPEG compression schemes) further improve compression rates by addressing a sequence of video frames and taking advantage of the temporal redundancy. Accordingly, compressed video includes target video elements (such as 8×8 blocks, slices, or frames) that may be predicted by a reference video element and additional information representative of the difference between the reference video element and the target video element. This prediction is also termed motion compensation.

For example, MPEG-2 standard defines three types of frames, I-frames, B-frames and P-frames. I-frames are independent in the sense that they include the entire information that is required for displaying a picture. A P-frame is decoded in response to information embedded within a previous frame, while a B-frames is decoded in response to information embedded within both a preceding and succeeding frame. The prediction is done in the picture domain on an 8×8 block basis. Each 8×8 target block is compared to the content of the reference frame (the previous frame in the case of the P-frame) to find the best matching group of 8×8 reference elements (e.g.—the reference block). The offset between each 8×8 target block and the reference block is embedded within a motion vector. It is noted that the reference block may not be aligned with the 8×8 blocks of the reference frame, and may intersect with up to four blocks (e.g.—reference blocks).

N. Merhav and V. Bhaskaran acknowledge that in order to perform various manipulations on compressed video streams there is a need to include motion compensation elements within blocks. They suggest an algorithm for motion compensation, at “A Fast Algorithm for DCT-Domain Motion compensation”, HP Labs Technical Reports, HPL-95-I7, 1995 and at U.S. Pat. No. 5,708,732. As noted above, the target block may intersect with multiple reference blocks. In the case of 8×8 blocks, the horizontal offset (as well as the vertical offset) between the target block and a reference blocks can vary between 0 and 7. The offset is embedded within the motion vector. Merhav and Bhaskaran suggest to implement the insertion of motion compensation elements by performing matrix multiplications, whereas some of the matrix are pre-calculated and are selected in response to the horizontal as well as the vertical offset between the target block and an reference block. It is noted that this algorithm is based upon multiple matrix multiplications and is relative complex and resource consuming.

The invention provides a method and an apparatus for generating a motion compensated matrix, and a computer readable medium having code embodied therein for causing an electronic device to perform the steps of: providing a motion vector and at least one input matrix of DCT elements; selecting at least one conversion transform (CT) and at least one inverse conversion transform (ICT) in response to a content of the motion vector; applying the at least one selected ICT and the at least one selected CT to provide a motion compensated matrix.

The invention provides an apparatus and method for generating a motion compensated matrix, the method includes the following steps: (i) receiving a motion vector and selecting a group of reference blocks that belong to a reference picture and determining a horizontal offset H and a vertical offset V between the members of the group and a target block; (ii) selecting a row wise ICT and a row wise CT in response to H; (iii) applying the selected row wise ICT on rows of two reference blocks to provide a first intermediate matrix; (iv) applying the selected row wise ICT on rows of other two reference blocks to provide a second intermediate matrix; (v) selecting a column wise ICT and a column wise CT in response to V; (vi) applying the selected column wise ICT on the columns of the first and second intermediate matrixes to provide a third intermediate matrix; (vii) applying the selected column wise CT on the columns of the third intermediate matrix to provide a fourth intermediate matrix; (viii) applying the selected row wise CT on the rows of the forth intermediate matrix to provide a motion compensated matrix. The method may also include an additional step of dividing the elements of the motion compensated by a normalization factor.

The invention provides an apparatus and method for generating a motion compensated matrix, including the steps of: receiving a motion vector and selecting a group of reference blocks that belong to a reference picture and determining a horizontal H and a vertical offset V between the members of the group and a target block; selecting a column wise ICT and a column wise CT in response to V; applying the selected column wise ICT on columns of two reference blocks to provide a first intermediate matrix; applying the selected column wise ICT on columns of other two reference blocks to provide a second intermediate matrix; selecting a row wise ICT and a row wise CT in response to H; applying the selected row wise ICT on the rows of the first and second intermediate matrixes to provide a third intermediate matrix; applying the selected row wise CT on the rows of the third intermediate matrix to provide a fourth intermediate matrix; applying the selected column wise CT on the column of the fourth intermediate matrix to provide a motion compensated matrix. The method may also include an additional step of dividing the elements of the motion compensated by a normalization factor.

Further features and advantages of the invention will be apparent from the description below. The invention is herein described, by way of example only, with reference to the accompanying drawings, wherein:

FIGS. 1 a – 1 c illustrate methods for motion compensation information enhancement, in accordance to various embodiments of the invention;

FIGS. 2 a and 2 b illustrate a first type ICT and a first type CT, in accordance to an aspect of the invention;

FIGS. 3 a and 3 b illustrate a second type ICT and a second type CT to be applied when an offset of one pixel is present, in accordance to an aspect of the invention;

FIGS. 4 a and 4 b illustrate a second type ICT and a second type CT to be applied when an offset of two pixels is present, in accordance to an aspect of the invention;

FIGS. 5 a and 5 b illustrate a second type ICT and a second type CT to be applied when an offset of three pixels is present, in accordance to an aspect of the invention;

FIGS. 6 a and 6 b illustrate a second type ICT and a second type CT to be applied when an offset of four pixels is present, in accordance to an aspect of the invention;

FIGS. 7 a and 7 b illustrate a second type ICT and a second type CT to be applied when an offset of five pixels is present, in accordance to an aspect of the invention;

FIGS. 8 a and 8 b illustrate, a second type ICT to be applied when an offset of six pixels is present, in accordance to an aspect of the invention;

FIGS. 9 a and 9 b illustrate a second type ICT and a second type CT to be applied when an offset of seven pixels is present, in accordance to an aspect of the invention; and

FIG. 10 illustrates an apparatus connected to multiple clients over a network, in accordance to an aspect of the invention.

According to various aspects of the invention the system and method for removing motion compensation information that utilize concise transforms such as a first type and second type conversion transform (CT) and inverse conversion transforms (ICT) perform cosine discrete operation and inverse cosine discrete operations but are characterized by fewer computations than prior art DCT and IDCT.

Each ICT if further operable to receive multiple input elements are to generate a fewer amount of output elements. The ICT are designed so as to omit calculations that are not required for the generation of these output elements.

Preferably, an ICT is operable to receive two input N×N matrixes and to output a single N×N matrix. Applying ICT and CT on DCT coefficient matrix provides a motion compensated matrix that includes motion compensation information and may be processed without requiring previous matrixes or preceding matrixes.

According to various aspects of the invention a first type of CT and ICT are presented wherein an amount of calculations is reduced by omitting calculations that cancel each other during a sequence that includes applying a ICT that is followed by applying a CT. Usually, calculations that belong to the last stage of a multi-stage first type ICT as well as reciprocal calculations that belong to the first stage of a multistage CT are omitted.

According to yet a further aspect of the invention a second type of CT and ICT are introduced. The second type CT and ICT are even more concise than the first type CT and ICT. The reduction results from the manner in which the ICT and CT are applied during the motion compensation information enhancement process. Usually, the ICT is applied more times than the CT. Accordingly, shifting some calculations from the ICT to the CT results in an overall reduction in the amount of calculations.

According to another aspect of the invention at least one pair of CT and ICT are tailored to each offset value.

According to yet a further aspect of the invention the CT and ICT are either two-dimensional of one-dimensional. Applying a row-wise CT and then applying a column-wise CT (or vice verse) is equivalent to an appliance of a two dimensional CT. The same follows for the ICT.

The invention provides a method for generating a motion compensated matrix, the method including the following steps: (i) providing a motion vector and at least one input matrix of DCT compressed elements; (ii) selecting at least one conversion transform (CT) and at least one inverse conversion transform (ICT) in response to a content of the motion vector, and (iii) applying the at least one selected ICT and the at least one selected CT to provide a motion compensated matrix.

For simplicity of explanation it is assumed that a row wise ICT is applied prior to a column wise ICT and that each blocks include 8×8 elements. A group of reference blocks includes an upper left reference block (M 1 ), an upper right reference block (M 2 ), a lower left reference block (M 3 ) and a lower right reference block (M 4 ). A vertical offset V is measured from the upper row of M 1 and M 2 to the upper row of the target block TB, and a horizontal offset H is measured from the most left-hand column of M 1 and M 3 to the most left-hand column of TB. The compressed representation of M 1 –M 4 and EB are denoted M 1 C–M 4 C and EBC accordingly.

First Embodiment of Motion Compensation Information Enhancement

FIGS. 1 a – 1 c illustrate methods for motion compensation information enhancement, in accordance to various embodiments of the invention.

Method 21 of FIG. 1 a starts by step 20 of receiving a motion vector and selecting a group of reference blocks (denoted M 1 C–M 4 C) that belong to a reference picture as well as determining a horizontal H and a vertical offset V between the members of the group (actually the upper left element of M 1 C) and the target block TC. Usually, the motion vector is associated with a certain block of a target picture that belongs to a compressed video stream. The motion vector defined an offset between that certain block and the reference block. The motion compensation information enhancement operates on a block to block basis, so that said method is repeated for all blocks of the target picture.

Step 20 is followed by step 22 of selecting a first type row wise ICT and a first type row wise CT in response to H. As illustrated below, for N=8 there are seven first type row wise ICT and CT, each best fits a certain H value.

Step 22 is followed by step 24 of applying the selected first type row wise ICT on the rows of M 1 C and M 2 C to provide a first intermediate matrix. It is noted that during each iteration out of the eight iterations of step 24 a single row of M 1 C as well as a corresponding single row of M 2 C are processed, to provide a single row of the first intermediate matrix.

Step 24 is followed by step 26 of applying the selected first type row wise ICT on the rows of M 3 C and M 4 C to provide a second intermediate matrix.

Step 26 is followed by step 28 of selecting a first type column wise ICT and a first type column wise CT in response to V. As illustrated below, for N=8 there are seven first type column wise ICT and CT, each best fits a certain V value.

Step 28 is followed by step 30 of applying the selected first type column wise ICT on the columns of the first and second intermediate matrixes to provide a third intermediate matrix. It is noted that during each iteration out of the eight iterations of step 30 a single column of the first intermediate matrix as well as a corresponding column of the second intermediate matrix are processed, to provide a single column of the third intermediate matrix.

Step 30 is followed by step 32 of applying the selected first type column wise CT on the columns of the third intermediate matrix to provide a fourth intermediate matrix.

Step 32 is followed by step 34 of applying the selected first type row wise CT on the rows of the third intermediate matrix to provide a motion compensated matrix.

FIGS. 1 b and 1 c illustrate that the order of some of the mentioned above steps may be changed (for example, applying a first type column wise ICT before applying a first type row wise ICT) and that second type CT and ICT may be implemented instead of first type CT and ICT.

For simplicity of explanation all the intermediate matrix are described by the same terms, throughout FIGS. 1 a – 1 c , although they differ from each other.

Method 121 of FIG. 1 b starts by step 120 of receiving a motion vector and selecting a group of reference blocks (denoted M 1 C–M 4 C) that belong to a reference picture as well as determining a horizontal H and a vertical offset V between the members of the group (actually the upper left element of M 1 C) and the target block TB.

Step 120 is followed by step 122 of selecting a second type row wise ICT and a second type row wise CT in response to H. As illustrated below, for N=8 there are seven second type row wise ICT and CT, each best fits a certain H value.

Step 122 is followed by step 124 of applying the selected second type row wise ICT on the rows of M 1 C and M 2 C to provide a first intermediate matrix. It is noted that during each iteration out of the eight iterations of step 124 a single row of M 1 C as well as a corresponding single row of M 2 C are processed, to provide a single row of the first intermediate matrix.

Step 124 is followed by step 126 of applying the selected second type row wise ICT on the rows of M 3 C and M 4 C to provide a second intermediate matrix.

Step 126 is followed by step 128 of selecting a second type column wise ICT and a second type column wise CT in response to V. As illustrated below, for N=8 there are seven second type column wise ICT and CT, each best fits a certain V value.

Step 128 is followed by step 130 of applying the selected second type column wise ICT on the columns of the first and second intermediate matrixes to provide a third intermediate matrix. It is noted that during each iteration out of the eight iterations of step 30 a single column of the first intermediate matrix as well as a corresponding column of the second intermediate matrix are processed, to provide a single column of the third intermediate matrix.

Step 130 is followed by step 132 of applying the selected second type column wise CT on the columns of the third intermediate matrix to provide a fourth intermediate matrix.

Step 132 is followed by step 134 of applying the selected second type row wise CT on the rows of the third intermediate matrix to provide a motion compensated matrix.

Method 221 of FIG. 1 c starts by step 220 of receiving a motion vector and selecting a group of reference blocks (denoted M 1 C–M 4 C) that belong to a reference picture as well as determining a horizontal H and a vertical offset V between the members of the group (actually the upper left element of M 1 C) and the target block TB.

Step 220 is followed by step 222 of selecting a second type column wise ICT and a second type column wise CT in response to V. As illustrated below, for N=8 there are seven second type column wise ICT and CT, each best fits a certain H value.

Step 222 is followed by step 224 of applying the selected second type column wise ICT on the columns of M 1 C and M 3 C to provide a first intermediate matrix. It is noted that during each iteration out of the eight iterations of step 24 a single column of M 1 C as well as a corresponding single column of M 3 C are processed, to provide a single column of the first intermediate matrix.

Step 224 is followed by step 226 of applying the selected second type column wise ICT on the columns of M 2 C and M 4 C to provide a second intermediate matrix.

Step 226 is followed by step 228 of selecting a second type row wise ICT and a second type row wise CT in response to H. As illustrated below, for N=8 there are seven second type row wise ICT and CT, each best fits a certain H value.

Step 228 is followed by step 230 of applying the selected second type row wise ICT on the rows of the first and second intermediate matrixes to provide a third intermediate matrix.

Step 230 is followed by step 232 of applying the selected second type row wise CT on the rows of the third intermediate matrix to provide a fourth intermediate matrix.

Step 232 is followed by step 234 of applying the selected second type column wise CT on the columns of the third intermediate matrix to provide a motion compensated matrix.

First Type CT and ICT Functions

The first type CT and ICT are made of various “building blocks”, but it is noted that various calculations are omitted from these “building blocks”, such as calculations that cancel each other and calculations that are not required for the provision of a final result. FIGS. 2 a and 2 b illustrate a first type ICT and CT for an offset of a single pixel, whereas the “building blocks” are illustrated by the following equations. Preferred combinations of building blocks are provided at table 1 . Those of skill in the art will appreciate how to combine the building block of table 1 for providing the first type CT and ICT, based upon the following description.

First Building Block

A 0= I 0+ I 7 ; A 1= I 1+ I 2 ; A 2 =I 1− I 2; A 3= I 3+ I 4 ; A 4 =I 3 −I 4 ; A 5= I 5+ I 6 ; A 6= I 5− I 6 A 7 =I 0− I 7.

B 0= A 0+ A 3; B 1= A 1+ A 5; B 2= A 2+ A 6; B 3= A 0− A 3; B 4=√{square root over (2)} *A 4; B 5= A 1− A 5; B 6= A 2− A 6; B 7=√{square root over (2)} *A 7.

C 0= B 0+ B 1; C 1= B 0− B 1; C 2= B 2+ B 4; C 3, C 6= ROT (√{square root over (2)} C− 6)[ B 3, B 6]; C 4= B 2− B 4; C 5= B 5+ B 7; C 7= B 5− B 7;

O0=C0 ; O 4=C1, O 1, O 7= ROT ( C 2)[ C 1, C 5]; O6=C3 ; O 3,− O 5= ROT ( C− 5)[ C 4, C 7], O 2= C 6.

Second Building Block

A 0= I 0+ I 7 ; A 1= I 1+ I 5 ; A 2= I 2+ I 6 ; A 3= I 3+ I 4 ; A 4= I 3 −I 4 ; A 5= I 1 −I 5 ; A 6= I 2 −I 6 ; A 7= I 0 −I 7.

B 0= A 0+ A 3; B 1= A 1+ A 2; B 2= A 1− A 2; B 3= A 0− A 3; B 4=√{square root over (2)} *A 4; B 5= A 5+ A 6; B 6= A 5− A 6; B 7=√{square root over (2)} *A 7;

C 0= B 0+ B 1; C 1= B 0− B 1; C 2= B 2+ B 4; C 3, C 6= ROT (√{square root over (2)} C− 6)[ B 3, B 6]; C 4= B 2− B 4; C 5= B 5+ B 7; C 7= B 5− B 7;

O0=C0; O4=C1, O 1, O 7= ROT ( C 2)[ C 1, C 5]; O 6= C 3; O 3,− O 5= ROT ( C− 5)[ C 4, C 7], O 2= C 6.

Third Building Block

A 0= I 0+ I 3 ; A 1= I 1+ I 6 ; A 2= I 2+ I 5 ; A 3= I 0− I 3; A 4= I 4+ I 7 ; A 5= I 2 −I 5 ; A 6= I 1 −I 6 ; A 7= I 4 −I 7.

B 0= A 0+ A 4; B 1= A 1+ A 2; B 2= A 1− A 2; B 3= A 3+ A 7; B 4= A 0− A 4; B 5=√{square root over (2)} *A 5; B 6=√{square root over (2)} *A 6; B 7= A 3− A 7;

C 0= B 0+ B 1; C 1= B 0− B 1; C 2, C 7= ROT (√{square root over (2)} C 6)[ B 2, B 7]; C 3= B 3+ B 5; C 4= B 6+ B 4; C 5= B 3− B 5; C 6= B 4− B 6.

O0=C0; O4=C1; O2=C2; O 1,− O 7= ROT ( C 5)[ C 3, C 4]; O 3, O 5= ROT ( C 1)[ C 5, C 6]; O6=C7.

Fourth Building Block

A0= I 0+ I 4 ; A 1= I 1+ I 6 ; A 2= I 2+ I 5 ; A 3= I 7+ I 3 ; A 4= I 0 −I 4 ; A 5= I 2 −I 5 ; A 6= I 1 −I 6 ; A 7= I 3 −I 7.

B 0= A 0+ A 3; B 1= A 1+ A 2; B 2= A 1− A 2; B 3= A 0− A 3; B 4= A 4+ A 7; B 5=√{square root over (2)} *A 5; B 6=√{square root over (2)} *A 6 ; B 7= A 4− A 7;

C 0= B 0+ B 1; C 1= B 0− B 1; C 2, C 7= ROT (√{square root over (2)} C 6)[ B 2, B 7]; C 3= B 3+ B 5; C 4= B 6+ B 4; C 5= B 3− B 5; C 6= B 4− B 6.

O0=C0; O4=C1; O2=C2; O 1 ,−O 7 =ROT ( C 5)[ C 3, C 4]; O 3, O 5= ROT ( C 1)[ C 5, C 6]; O6=C7.

Fifth Building Block

A 0= I 0+ I 7; A 1= I 1+ I 6 ; A 2= I 2+ I 5 ; A 3= I 3+ I 4 ; A 4= I 3 −I 4 ; A 5= I 2 −I 5 ;A 6= I 1 −I 6;

A 7= I 0− I 7.

B 0= A 0+ A 3; B 1= A 1+ A 2; B 2= A 1− A 2; B 3= A 0− B 3; B 4, B 7= ROT ( C 3)[ A 4, A 7]; B 5, B 6= ROT ( C 3)[ A 5, A 7];

C 0= B 0+ B 1; C 1= B 0− B 1; C 2, C 3= ROT (√{square root over (2)} C 1)[ B 2, B 3]; C 4= B 4+ B 6; C 5= B 7− B 5; C 6= B 4− B 6; C 7= B 5+ B 7;

O0=C0; O 1= C 7+ C 4; O 2= C 2; O 3=√{square root over (2)} C 5; O4=C1; O 5 √{square root over (2)} C 6; O6=C3; O 7= C 7 −C 4

First Type ICT and CT for an Offset of One Pixel.

FIGS. 2 a and 2 b illustrate a first type ICT and a first type CT, in accordance to an aspect of the invention.

The following example illustrates a first type row wise ICT and CT, although the same applies to first type column wise CT and ICT.

Addition operation, subtraction operation and ROT operations are graphically illustrated by the following signs:

Addition of a first variable with a second variable: a straight line extending from the first variable that is connected by a dot to another straight line extending from the second variable.

Subtraction of a first variable from a second variable: a dashed line extending from the first variable that is connected by a dot to another straight line extending from the second variable.

Rotation operation (ROT) is denoted by a rectangle having two inputs and two outputs. A multiplication by √{square root over (2)} is denoted by “0”, a multiplication by 2 is denoted by “<<” and a multiplication by 2*√{square root over (2)} is denoted by “0<<”.

FIG. 2 a illustrated the ICT operation. The upper portion of FIG. 2 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0 ; A 1, A 7= ROT ( C− 5)[ I 1,− I 7]; A 2=I2 ; A 3, A 5= ROT ( C− 1)[I3, I 5], A4=I4 and A6=I6. B0=A0; B 1= A 1+ A 3; B 2, B 6= ROT (√{square root over (2)} C− 6)[ A 2, A 6]; B 3= A 1− A 3; B 4= A 0− A 4; B 5= A 7− A 5; B 7= A 7+ A 5. C0=B0; C 2= B 4− B 2; C 3=√{square root over (2)} B 3; C 4= B 4+ B 2; C 5=√{square root over (2)} *B 5; C 6= B 7− B 6; C 7= B 7+ B 6. IR 0= C 5+ C 4; IR 1= C 2+ C 3; IR 2= C 1+ C 6; IR 3= C 0 −C 7 ; IR 4= C 2 −C 3 ; IR 5= C 4 −C 5; IR 6= C 1 −C 6.

The lower portion of FIG. 2 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A2=I2; A4=I4; A 5=LOW_ROT(√{square root over (2)} C 5)[ I 3 ,−I 5]; A6=I6 ; A 7=LOW_ROT(√{square root over (2)} C− 7)[I1, I 7].

B 0= A 0+ A 4; B 5= A 7− A 5; B 6=UP_ROT(√{square root over (2)} C 6)[ A 6, A 2];

C 0= B 0+ B 6; C5=B5

IR 7= C 0+ C 5.

FIG. 2 b illustrates a CT that is applied on a matrix. It is also defined by the following set of equations:

A 0= I 0 +I 5 ; A 1 =I 2 +I 1 ; A 2 =I 1 −I 2 ; A 3 =I 6 −I 3 ; A 4 =I 7 +I 4; A 5= I 0− I 5; A 6= I 3+ I 6 ; A 7= I 4 −I 7; B 0= A 0+ A 6; B 1= A 4− A 1; B 2= A 7− A 2; B 3=√{square root over (2)} A 3; B 4= A 1+ A 4; B 5=√{square root over (2)} *A 5; B 6= A 0− A 6; B 7= A 2+ A 7; C 0= B 0+ B 4; C 1= B 1+ B 3; C 6, C 2= ROT (√{square root over (2)} C− 6)[ B 6, B 2]; C 3= B 1− B 3; C 4= B 0− B 4; C 5= B 7− B 5; C 7= B 5+ B 7; O0=C0 ; O 1, O 7= ROT ( C 7)[ C 1, C 7); O2=C2; O 3,− O 5= ROT ( C− 5)[ C 3, C 5], O4=C4; O6=C6.

Second Type ICT and CT for an Offset of One Pixel.

The upper portion of FIG. 3 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2; A 3, A 4= ROT ( C− 5)[I1 ,−I 7 ]; A 5, A 6= ROT ( C− 1)[I3 , I 5] and A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT ((√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 5+ A 3; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 0= B 0+ B 3; C1=B1; C2=B2; C 3= B 0− B 3; C 4= B 4+ B 7; C5=B5; C6=B6 and C 7 =B 4− B 7.

O0=C1; O1=C2; O=C3; O 3= C 0 −C 4; O4=C5, O5=C6, and O6=C7.

The lower portion of FIG. 3 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A 5=LOW_ROT(√{square root over (2)} C− 7)[ I 1, I 7]; A3=I6; A6=I2 ; A 7=LOW_ROT(√{square root over (2)} C 5)[ I 3 ,−I 5].

B 0= A 0+ A 1; B 3=HIGH-ROT(√{square root over (2)} C 6)[ A 3, A 6]; B 7= A 5− A 7;

C 0= B 0+ B 3; C7=B7.

O 7= C 0+ C 7.

FIG. 3 b illustrates a CT that is applied on a matrix. It is also defined by the following set of equations:

A 0= I 0+ I 1 ; A 1= I 0− I 1; A2=I2; A3=I3 ; A 4=√{square root over (2)} *I 4; A 5=√{square root over (2)} *I 5; A6=I6; A7=I7. B 0= A 0+ A 5; B 1= A 1+ A 4; B2=A2; B3=A3; B 4= A 1− A 4; B 5= A 0− A 5; B6=A6; B7=A7. C 0= B 0+ B 7; C 1= B 1+ B 5; C 2=2 *B 2; C 3= B 3+ B 4; C 4= B 3− B 4; C 5= B 1− B 5; C 6=2 *B 6; C 7= B 0− B 7. D 0= C 0+ C 3; D 1= C 1+ C 2; D 2= C 1 −C 2 ; D 3= C 0 −C 3 ; D 4=√{square root over (2)} *C 4; D 5= C 5+ C 6; D 6= C 5 −C 6 ; D 7=√{square root over (2)} *C 7. E 0= D 0+ D 1; E 1= D 0− D 1; E 2= D 2+ D 4; E 3, E 6= ROT (√{square root over (2)} E− 6)[ D 3, D 6]; E 4= D 2− D 4; E 5= D 5+ D 7; E 7= D 5− D 7; O0=E0 ; O 4= E 1; O 1, O 7= ROT ( C 7)[ E 2, E 5); O6=E3; O 3, − O 5= ROT ( C− 5)[ E 4, E 7]; O2=E6.

Second Type ICT and CT for an Offset of Two Pixels.

The upper portion of FIG. 4 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2 ; A 3, A 4= ROT ( C− 5)[ I 1 ,−I 7 ]; A 5, A 6= ROT ( C− 1)[ I 3, I 5], A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT ((√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 5+ A 3; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 0= B 0+ B 4; C 1= B 1+ B 2; C 2= B 1− B 2; C 3= B 3+ B 7; C 4= B 0− B 4; C5=B5 ; C 6=√{square root over (2)} B 6 and C 7= B 3− B 7.

O0=C2 ; O 1= C 0 −C 3; O2=C4; O3=C5; O 4= C 1 −C 6 and O5=C7.

The lower portion of FIG. 4 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2; A3=I6 ; A 4= I 1 −I 7 ; A 5=√{square root over (2)} *I 3; A 6=√{square root over (2)} *I 5 ; A 7= I 7+ I 1.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 3= ROT (√{square root over (2)} C− 6)[ A 2, A 3]; B 4= A 4+ A 6; B 5= A 7− A 5; B 6= A 4− A 6; B 7= A 5+ A 7.

C 0= B 0+ B 3; C 1= B 1+ B 2; C 6=LOW_ROT( C− 1)[ B 5, B 6], C 7=LOW_ROT( C− 3)[ B 4, B 7].

O 6 =C 0+ C 7; O 7= C 1+ C 6.

FIG. 4 b illustrates a CT that is applied on a matrix. It is also defined by the following set of equations:

A 0=2 *I 0; A1=I1 ; A 2=2* I 2 ; A 3=2√{square root over (2)} *I 3 ; A 4= I 4+ I 7 ; A 5=2√{square root over (2)} *I 5; A 6= I 1 −I 6 ; A 7= I 4 −I 7. B 0= A 0+ A 4; B 1= A 1+ A 2; B 2= A 1− A 2; B 3= A 3+ A 7; B 4= A 0− A 4; B 5= A 5; B 6=√{square root over (2)} *A 6; B 7= A 3− A 7. C 0= B 0+ B 1; C 1= B 0− B 1; C 2, C 7= ROT (√{square root over (2)} C 6)[ B 2, B 7]; C 3= B 5+ B 3; C 4= B 4+ B 6; C 5= B 3− B 5; C 6= B 4− B 6. O0=C0; O4=C1; O2=C2 ; O 1,− O 7= ROT ( C 5)[ C 3, C 4]; O 3, O 5= ROT ( C 1)[ C 5, C 6]; O6=C7.

Second Type ICT and CT for an Offset of Three Pixels

The upper portion of FIG. 5 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2 ; A 3, A 4= ROT ( C− 5)[ I 1 ,−I 7 ]; A 5, A 6= ROT ( C− 1)[ I 3, I 5] and A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT ((√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 5+ A 3; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 0= B 0+ B 3; C 1= B 1+ B 2; C 2= B 1− B 2; C 3= B 0− B 3; C 4= B 4+ B 7; C 5=√{square root over (2)} *B 5; C 6=√{square root over (2)} *B 6 and C 7= B 4− B 7.

O0=C3; O 1= C 0 −C 4; O 2 =C 2 −C 5; O 3 =C 1 −C 6; O4=C7.

The lower portion of FIG. 5 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A 2, A 5= ROT ( C− 7)[I1, I 7]; A3=I6 ; A 4, A 7= ROT ( C 5)[ I 3 ,−I 5]; A6=I2.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2= A 2+ A 4; B 3, B 6= ROT (√{square root over (2)} C 6)[ A 3, A 6]; B 5= A 5+ A 7; B 7= A 5− A 7.

C 0= B 0+ B 3; C 1= B 1+ B 5; C 2= B 2+ B 6; C 7=√{square root over (2)} *B 7.

O 5 =C 0+ C 7; O6=C1; O7=C2.

FIG. 5 b illustrates a CT that is applied on a matrix. It is also defined by the following set of equations:

A0=I0; A1=I1; A2=I2; A3=I3; A4=I4; A5=I5; A6 =I 6+ I 7 ; A 7= I 6− I 7. B 0=2 *A 0; B 1= A 1+ A 6; B 2= A 5+ A 2; B 3= A 3+ A 7; B 4=2 *A 4; B 5= A 2− A 5; B 6= A 1− A 6; B 7= A 3− A 7. C 0= B 0+ B 3; C 1= B 1+ B 2; C 2= B 1− B 2; C 3= B 0− B 3; C 4= B 4+ B 7; C 5=√{square root over (2)} *B 5; C 6=2* B 6; C 7= B 4− B 7. D 0= C 0+ C 1; D 1= C 0− C 1; D 2, D 7= ROT (√{square root over (2)} C 6)[ C 2, C 7]; D 3= C 3+ C 5; D 4= C 4+ C 6; D 5= C 3 −C 5 ; D 6= C 4 −C 6. O0=D0; O4=D1; O2=D2; O 1,− O 7= ROT ( C 5)[ D 3, D 4]; O 3, O 5= ROT ( C 1)[ D 5, D 6]; O6=D7.

Second Type ICT and CT for an Offset of Four Pixels.

The upper portion of FIG. 6 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A2=I2; A2=I2 ; A 3, A 4= ROT ( C− 5)[ I 1 ,−I 7 ]; A 5, A 6= ROT ( C− 1)[I3, I 5] and A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT (√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 3+ A 5; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 1= B 1+ B 2; C 2= B 1− B 1; C 4= B 0− B 4; C 5=√{square root over (2)} *B 5; C 6=√{square root over (2)} *B 6; C 7= B 3− B 7.

O0=C4 ; O 1 =C 2 −C 5 ; O 2 =C 1− C 6; O3=C7.

The lower portion of FIG. 6 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2 ; A 3, A 4= ROT ( C− 5)[ I 1 ,−I 7 ]; A 5, A 6= ROT ( C− 1)[ I 3, I 5] and A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT ((√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 5+ A 3; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 0= B 0+ B 4; C 1= B 1+ B 2; C 2= B 1− B 2; C 3= B 3+ B 7; C 5=√{square root over (2)} *B 5; C 6=√{square root over (2)} *B 6.

O4=C0 ; O 5 =C 1+ C 6; O 6= C 2+ C 5; O7=C3.

FIG. 6 b illustrates a CT that is applied on a matrix. It is also defined by the following set of equations:

A 0=2 *I 0; A 1= I 1+ I 6 ; A 2= I 2+ I 5 ; A 3=2* I 3 ; A 4=2* I 4 ; A 5= I 2 −I 5 ; A 6= I 1 −I 6 ; A 7=2* I 7. B 0= A 0+ A 4; B 1= A 1+ A 2; B 2= A 1− A 2; B 3= A 3+ A 7; B 4= A 0− A 4; B 5=√{square root over (2)} *A 5; B 6=√{square root over (2)} *A 6; B 7= A 3− A 7. C 0= B 0+ B 1; C 1= B 0− B 1; C 2, C 7= ROT (√{square root over (2)} C 6)[ B 2, B 7]; C 3= B 3+ B 5; C 4= B 4+ B 6; C 5= B 3− B 5; C 6= B 4− B 6. O0=C0; O4=C1; O2=C2; O 1,− O 7= ROT ( C 5)[C3, C 4]; O 3, O 5 =ROT ( C 1)[ C 5, C 6]; O6=C7.

Second Type ICT and CT for an Offset of Five Pixels

The upper portion of FIG. 7 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4 ; A 2, A=ROT ( C− 7)[I1, I 7 ]; A 3= I 6 ; A 4, A 7= ROT ( C 5)[ I 3 ,−I 5] and A6=I2.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2= A 2+ A 4; B 3, B 6= ROT ((√{square root over (2)} C− 6)[ A 3, A 6]; B 5= A 5+ A 7; B 7= A 5− A 7.

C 0= B 0+ B 3; C 5= B 1− B 5; C 6= B 2− B 6; C 7=√{square root over (2)} *B 7.

O0=C5; O1=C6 ; O 2= C 0 −C 7.

The lower portion of FIG. 7 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2 ; A 3, A 4= ROT ( C− 5)[I1 ,−I 7 ]; A 5, A 6= ROT ( C− 1)[ I 3, I 5] and A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT ((√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 5+ A 3; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 0= B 0+ B 3; C 1= B 1+ B 2; C 2= B 1− B 2; C 3= B 0− B 3; C 4= B 4+ B 7; C 5=√{square root over (2)} *B 5; C 6=√{square root over (2)} *B 6 and C 7= B 4− B 7.

O3=C0 ; O 4= C 1+ C 6; O 5= C 2+ C 5; O 6= C 3+ C 7; O7=C4.

FIG. 7 b illustrates a CT that is applied on a matrix. It is also defined by the following set of equations:

A 0= I 0+ I 1 ; A 1= I 0 −I 1 ; A 2=I2; A3=I3; A4=I4; A5=I5; A6=I6; A7=I7. B 0= A 0+ A 4; B 1= A 1+ A 6; B 2= A 5+ A 2; B 3=2 *A 3; B 4= A 0− A 4; B 5= A 2− A 5; B 6= A 1− A 6; B 7=2*. C 0= B 0+ B 3; C 1= B 1+ B 2; C 2= B 1− B 2; C 3= B 0− B 3; C 4= B 4+ B 7; C 5=√{square root over (2)} *B 5; C 6=√{square root over (2)} *B 6; C 7= B 4− B 7. D 0= C 0+ C 1; D 1= C 0 −C 1; D 2, D 7= ROT (√{square root over (2)} C 6)[ C 2, C 7]; D 3= C 3+ C 5; D 4= C 4+ C 6; D 5= C 3 −C 5 ; D 6= C 4 −C 6. O0=D0; O4=D1; O2=D2 ; O 1 ,−O 7= ROT ( C 5)[ D 3, D 4]; O 3, O 5= ROT ( C 1)[ D 5, D 6]; O6=D7.

Second Type ICT and CT for an Offset of Six Pixels.

The upper portion of FIG. 8 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2; A3=I6 ; A 4= I 1 −I 7 ; A 5=√{square root over (2)} *I 3; A 6=√{square root over (2)} *I 5; A 7= I 7+ I 1.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 3= ROT (√{square root over (2)} C− 6)[ A 2, A 3]; B 4= A 4+ A 6; B 5= A 7− A 5; B 6= A 4− A 6; B 7= A 5+ A 7.

C 0= B 0+ B 3; C 1= B 1+ B 2; C 6=LOW_ROT( C− 1)[ B 5, B 6], C 7=LOW_ROT( C− 3)[ B 4, B 7].

O 0 =C 1 −C 6; O 1 =C 0 −C 7.

The lower portion of FIG. 8 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2 ; A 3, A 4= ROT ( C− 5)[ I 1 ,−I 7 ]; A 5, A 6= ROT ( C− 1)[ I 3, I 5], A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT ((√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 5+ A 3; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 0= B 0+ B 4; C 1= B 1+ B 2; C 2= B 1− B 2; C 3= B 3+ B 7; C 4= B 0− B 4; C 5= B 5; C 6=√{square root over (2)} *B 6 and C 7= B 3− B 7.

O2=C0; O 3= C 1+ C 6; O4=C2; O5=C3; O 6= C 4+ C 7 and O7=C5.

FIG. 8 b illustrates a CT that is applied on a matrix is defined by the following set of equations:

A 0= I 3+ I 0 ; A 1= I 1+ I 6 ; A 2=2* I 2 ; A 3= I 0 −I 3 ; A 4=2* I 4 ; A 5=2√{square root over (2)} *I 5; A 6= I 1 −I 6 ; A 7=2√{square root over (2)} *I 7. B 0= A 0+ A 4; B 1= A 1+ A 2; B 2= A 1− A 2; B 3= A 3+ A 7; B 4= A 0− A 4; B 5= A 5; B 6=√{square root over (2)} *A 6; B 7= A 3− A 7. C 0= B 0+ B 1; C 1= B 0− B 1; C 2, C 7= ROT ( C 6)[ B 2, B 7 ]; C 3= B 5+ B 3; C 4= B 4+ B 6; C 5= B 3− B 5; C 6= B 4− B 6. O0=C0; O4=C1; O2=C2 ; O 1,− O 7= ROT ( C 5)[ C 3, C 4]; O 3, O 5= ROT ( C 1)[ C 5, C 6]; O6=C7.

Second Type ICT and CT for an Offset of Seven Pixels.

The upper portion of FIG. 9 a illustrated a portion of the ICT that is applied on the upper matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4 ; A 5LOW_ROT(√{square root over (2)} C− 7)[ I 1, I 7]; A3=I6; A6=I2 ; A 7=LOW_ROT(√{square root over (2)} C 5)[ I 3 ,−I 5].

B 0= A 0+ A 1; B 3=HIGH_ROT(√{square root over (2)} C 6)[ A 3, A 6]; B 7= A 5− A 7;

C0=B0 ; C 7= B 3− B 7;

O 7= C 0+ C 7.

The lower portion of FIG. 9 a illustrated a portion of the ICT that is applied on the lower matrix (in the case of a row wise ICT). It is also defined by the following set of equations:

A0=I0; A1=I4; A2=I2 ; A 3, A 4= ROT ( C− 5)[ I 1 ,−I 7];

A 5, A 6= ROT ( C− 1)[ I 3, I 5] and A7=I6.

B 0= A 0+ A 1; B 1= A 0− A 1; B 2, B 7= ROT ((√{square root over (2)} C− 6)[ A 2, A 7]; B 3= A 5+ A 3; B 4= A 4+ A 6; B 5= A 3− A 5; B 6= A 4− A 6.

C 0= B 0+ B 3; C 1= B 1; C 2= B 2; C 3= B 0− B 3; C 4= B 4+ B 7; C5=B5; C6=B6 and C 7 =B 4− B 7.

O1=C0; O2=C1; O3=C2; O 4= C 3+ C 7; O5=C4, O6=C5, and O7=C6.

FIG. 9 b illustrates a CT that is applied on a matrix. It is also defined by the following set of equations:

A0=I0; A1=I1 ; A 2= I 2+ I 3 ; A 3= I 2 −I 3; A4=I4; A5=I5 ; A 6=√{square root over (2)} *I 6; A 7=√{square root over (2)} *I 7. B0=A0; B1=A1; B 2= A 2+ A 7; B 3= A 3+ A 6; B4=A4; B5=A5; B 6= A 3− A 6; B 7= A 2− A 7. C 0= B 0+ B 7; C 1=2 *B 1; C 2= B 6+ B 2; C 3= B 3+ B 4; C 4= B 3− B 4; C 5=2 *B 5; C 6= B 2− B 6; C 7= B 0− B 7. D 0= C 0+ C 3; D 1= C 1+ C 2; D 2= C 1 −C 2 ; D 3= C 0 −C 3 ; D 4=√{square root over (2)} *C 4; D 5= C 5+ C 6; D 6= C 5 −C 6 ; D 7=√{square root over (2)} *C 7. E 0= D 0+ D 1; E 1= D 0− D 1; E 2= D 2+ D 4; E 3, E 6= ROT (√{square root over (2)} E 6)[ D 3, D 6]; E 4= D 2− D 4; E 5= D 5+ D 7; E 7= D 5− D 7; O0=E0; O4=E1; O 1, O 7= ROT (C7)[ E 2, E 5]; O6=E3; O 3 , −O 5= ROT ( C− 5)[ E 4, E 7]; O2=E6.

Apparatus for Providing Motion Compensated Video Stream

FIG. 10 illustrates an apparatus 100 connected to multiple clients 170 over a network 180 , in accordance to an aspect of the invention.

System 100 includes first media processor 110 , storage unit 120 , second media processor 130 (also termed apparatus), transmitter 160 and packet generator 145 . Packet generator 145 preferably includes parser 140 and packetizer 150 . Transmitter 160 is connected to multiple clients 170 over network 180 . Packet generator 145 and transmitter may be included within a streaming server.

First media processor 110 is operable to receive video data, such as raw video stream, and to compress in according to a predefined compression scheme that includes DCT conversion. Conveniently, the compressed video is compressed according to standard compression schemes, such as MPEG 2, MPEG 4 and the like. provide standard compliant media streams and meta-data, such as MPEG 4 compliant media atoms and meta-data atoms.

First media processor 110 stores the compressed video streams at storage unit 120 . Second media processor 130 is able to fetch the video streams and perform motion compensation information enhancement. Assuming that the compressed video is compressed according to standard compression schemes, such as but not limited to MPEG 2 or MPEG 4, the second media processor is operable to fetch a motion vector, extract H and V, select and fetch a group of reference blocks and a target block and to execute at least one of the methods mentioned above.

Apparatus controller 220 is connected to first media processor 110 and media storage unit 120 in order to manage the storage of a motion compensated video stream, but this is not necessarily so and in many cases the motion compensated version may replace the compressed video version.

The motion compensated version as well as the compressed video may be further manipulated by system 101 in manners known in the art to provide a video stream or multiplex of video streams to a client.

Parser 140 , packetizer 150 and transmitter 160 co-operate to fetch, packetize and transmit video and audio (media) packets over network 180 to clients 170 . Network 180 may include wireless network, cable network, and various networks and combination of network elements.

It will be appreciated that the preferred embodiments described above are cited by way of example, and that the present invention is not limited to what has been particularly shown and described hereinabove. Rather, the scope of the present invention includes both combinations and sub-combinations of the various features described hereinabove, as well as variations and modifications thereof which would occur to persons skilled in the art upon reading the foregoing description and which are not disclosed in the prior art.