High dynamic range image information hiding method

The present invention claims priority under 35 U.S.C. 119(a-d) to CN 201710885604.2, filed Sep. 27, 2017.

The present invention relates to an image information hiding technology, and more particularly to a high dynamic range image information hiding method.

In the field of image processing, the dynamic range refers to the logarithm of the maximum and minimum luminosity ratios in a digital image. The dynamic range of the real scene reaches 9 orders of magnitude, but the existing low dynamic range (LDR) image adopts 8 bits/color/pixel to store data and loses part of the scene information. The high dynamic range (HDR) imaging technology makes up for the deficiency, which accurately shows the brightness range of the realistic scene through floating-point data, so that it is increasingly concerned and used in digital photography, ultra-high-definition movies and television, video games, remote sensing, medical imaging and other aspects.

Information hiding is a way to use digital media to hide important information. In recent years, the information hiding technology for low dynamic range images has developed rapidly, but researches on high dynamic range images have just started. The existing high dynamic range image hiding algorithm is able to be divided into two categories, wherein the first category sacrifices the image quality in exchange for higher embedding rate, the embedding rate of the existing top algorithm is only higher than 5 bits per pixel (5 bpps); the second category is able to achieve lossless embedding of information, but the embedding rate is less than 0.2 bits per pixel (0.2 bpp). Therefore, it is necessary to study a high dynamic range image information hiding method which is able to realize the balance between the embedding rate and the image quality.

A technical problem to be solved of the present invention is to provide a high dynamic range image information hiding method with high capacity and high fidelity, which is able to embed secret information into high dynamic range images in OpenEXR format, so that when an embedding rate is 30 bits per pixel, the image embedded with the secret information is still able to remain high fidelity.

Accordingly, in order to solve the above technical problem, the present invention adopts a technical solution as follows. A high dynamic range image information hiding method comprises steps of embedding secret information and extracting the secret information, wherein:

the step of embedding secret information comprises:

{circle around (1)}_1 recording an original high dynamic range image in OpenEXR format to be embedded with the secret information as I org , wherein: a height of the I org is R and a width thereof is C,

recording an original secret information sequence for being embedded into the original high dynamic range image as W, wherein: a value of every information of the W is an integer from 0 to 9, encrypting the W through encryption algorithm, obtaining an encrypted secret information sequence, recording the encrypted secret information sequence as W hide , and taking the encryption algorithm as a secret key Key 1 , wherein: a length of the W is equal to R×C×3;

{circle around (1)}_2 according to a numerical conversion formula in OpenEXR format, obtaining three channel values of every pixel in the I org , recording an i th channel value of a pixel whose coordinates are (x,y) in the I org as V i (x,y), wherein

V i ⁡ ( x , y ) = { ( - 1 ) S i ⁡ ( x , y ) ⁢ 2 - 14 ⁢ ( 0 + M i ⁡ ( x , y ) 1024 ) E i ⁡ ( x , y ) = 0 , M i ⁡ ( x , y ) > 0 ( - 1 ) S i ⁡ ( x , y ) ⁢ 2 E i ⁡ ( x , y ) - 15 ⁢ ( 1 + M i ⁡ ( x , y ) 1024 ) 1 ≤ E i ⁡ ( x , y ) ≤ 30 ,

here, 1≤x≤R, 1≤y≤C, i is equal to 1, 2 or 3, a first channel value represents R channel value, a second channel value represents G channel value, a third channel value represents B channel value, S i (x,y) represents one sign bit of the i th channel of the pixel whose coordinates are (x,y) in the I org , E i (x,y) represents 5-bit exponent of the i th channel of the pixel whose coordinates are (x,y) in the I org , M i (x,y) represents 10-bit mantissa of the i th channel of the pixel whose coordinates are (x,y) in the I org ;

{circle around (1)}_3 according to every channel value and corresponding 5-bit exponent of every pixel in the I org , determining an embedding significance bit of the information to be embedded in every channel value of every pixel in the I org , recording the embedding significance bit of the information to be embedded in the i th channel value of the pixel whose coordinates are (x,y) in the I org as ESB i (x,y), and recording the embedding significance bit of the information to be embedded in every channel value of every pixel in the I org as a secret key Key 2 ;

{circle around (1)}_4 embedding information into every channel value of every pixel in the I org , wherein: embedding information into the i th channel value V i (x,y) of the pixel whose coordinates are (x,y) in the I org comprises:

{circle around (1)}_4a recording a value of an ESB i (x,y) th embedding significance bit of the information to be embedded of the V i (x,y) as D ESB i (x,y), setting a j th information in the W hide as a current information to be embedded in the W hide , recording the j th information as D hide (j), wherein: D ESB i (x,y)∈[0,9], an initial value of the j is 1, 1≤j≤J, the J represents a length of the W, D hide (j)∈[0,9];

{circle around (1)}_4b judging whether the D ESB i (x,y) is equal to the D hide (j), wherein: if the D ESB i (x,y) is equal to the D hide (j), the V i (x,y) is kept unchanged, so that embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org is completed; if the D ESB i (x,y) is not equal to the D hide (j), the V i (x,y) is amended to complete embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org , wherein: the amendment comprises:

(1) setting V i,new (x,y)=V i (x,y)+N×SL i (x,y), (2) after the step (1), according to the V i,new (x,y), determining 5-bit exponent corresponding to the V i,new (x,y) and recording the 5-bit exponent as E i,new (x,y), and then through a manner as same as the step {circle around (1)}_3, according to the V i,new (x,y) and the E i,new (x,y), determining an embedding significance bit of the information to be embedded of the V i (x,y) and recording the embedding significance bit as ESB i,new (x,y); and then judging whether the D ESB i,new (x,y) of the ESB i,new (x,y) th embedding significance bit of the information to be embedded of the V i,new (x,y) is equal to the D hide (j), wherein if the D ESB i,new (x,y) is equal to the D hide (j), V i (x,y)=V i,new (x,y) is set, so that embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org is completed; if the D ESB i,new (x,y) is not equal to the D hide (j), a step (3) is executed; (3) setting V i,new (x,y)=V i (x,y)−N×SL i (x,y); (4) after the step (3), according to the V i,new (x,y), determining 5-bit exponent corresponding to the V i,new (x,y) and recording the 5-bit exponent as E i,new (x,y); and then through a manner as same as the step {circle around (1)}_3, according to the V i,new (x,y) and the E i,new (x,y), determining an embedding significance bit of the information to be embedded of the V i,new (x,y) and recording the embedding significance bit as the ESB i,new (x,y); and then judging whether the D ESB i,new (x,y) of the ESB i,new (x,y) th embedding significance bit of the information to be embedded of the V i,new (x,y) is equal to the D hide (j), wherein if the D ESB i,new (x,y) is equal to the D hide (j), V i (x,y)=V i,new (x,y) is set, so that embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org is completed; if the D ESB i,new (x,y) is not equal to the D hide (j), a step (5) is executed; and (5) setting N=N+1, and then returning to the step (1) to go on, wherein: the V i,new (x,y) is an introduced intermediate variable, N is an integer whose initial value is equal to 1, SL i (x,y) is a step length corresponding to the E i (x,y), and “=” in the SL i (x,y)=2 E i (x,y)−15 , V i (x,y)=V i,new (x,y) and N=N+1 is an assignment symbol; and

{circle around (1)}_5 according to the numerical conversion formula in OpenEXR format, storing every channel value embedded with the information of every pixel in the I org , and obtaining a high dynamic range image embedded with the secret information;

the step of extracting the secret information comprises:

{circle around (2)}_1 recording the high dynamic range image embedded with the secret information as I stego , wherein: a height of the I stego is R and a width thereof is C;

{circle around (2)}_2 according to the numerical conversion formula in OpenEXR format, obtaining three channel values embedded with the information of every pixel in the I stego , recording an i th channel value embedded with the information of a pixel whose coordinates are (x,y) in the I stego as V i ′(x,y), wherein:

V i ′ ⁡ ( x , y ) = { ( - 1 ) S i ′ ⁡ ( x , y ) ⁢ 2 - 14 ⁢ ( 0 + M i ′ ⁡ ( x , y ) 1024 ) E i ′ ⁡ ( x , y ) = 0 , M i ′ ⁡ ( x , y ) > 0 ( - 1 ) S i ′ ⁡ ( x , y ) ⁢ 2 E i ′ ⁡ ( x , y ) - 15 ⁢ ( 1 + M i ′ ⁡ ( x , y ) 1024 ) 1 ≤ E i ′ ⁡ ( x , y ) ≤ 30 ,

here, 1≤x≤R, 1≤y≤C, i is equal to 1, 2 or 3, a first channel value represents R channel value, a second channel value represents G channel value, a third channel value represents B channel value, S i ′(x,y) represents one sign bit of the i th channel of the pixel whose coordinates are (x,y) in the I stego , E i ′(x,y) represents 5-bit exponent of the i th channel of the pixel whose coordinates are (x,y) in the I stego , M i ′(x,y) represents 10-bit mantissa of the i th channel of the pixel whose coordinates are (x,y) in the I stego ;

{circle around (2)}_3 according to the secret key Key 2 , obtaining a position where the information is embedded of every channel value embedded with the information of every pixel in the I stego , and recording the position in the V i ′(x,y) as ESB i ′(x,y);

{circle around (2)}_4 extracting information from every channel value embedded with the information of every pixel in the I stego , wherein: extracting information from the V i ′(x,y) comprises: recording a value of an ESB i ′(x,y) th embedded position in the V i ′(x,y) as D ESB i ′ (x,y), and then taking the D ESB i ′ (x,y) as the information extracted from the V i ′(x,y) and recording as D ex (u), wherein: D ESB i ′ (x,y)∈[0,9], an initial value of the u is 1, a step length thereof is 1, D ex (u)∈[0,9]; and

{circle around (2)}_5 forming sequences through the information extracted from all channel values embedded with information of all pixels in the I stego , recording the sequences as W ex , decrypting the W ex according to the secret key Key 1 , obtaining decrypted secret information sequences, and recording the obtained decrypted secret information sequences as W*.

In the step {circle around (1)}_1, the encryption algorithm is Arnold transform or chaotic transformation, the W is scrambled through Arnold transform or chaotic transformation.

In the step 1_3, if V i (x,y)∈[2 E i (x,y)−15 ,2 E i (x,y)−14 ) and E i (x,y)∈[0,19], then ESB i (x,y)=5; if V i (x,y)∈[2 5 ,2 6 ) and E i (x,y)=20, then ESB i (x,y)=4; if V i (x,y)∈[2 6 ,10 2 ) and E i (x,y)=21, then ESB i (x,y)=3; if V i (x,y)∈[10 2 ,2 7 ) and E i (x,y)=21, then ESB i (x,y)=4; if V i (x,y)∈[2 7 ,2 8 ) and E i (x,y)=22, then ESB i (x,y)=3; if V i (x,y)∈[2 8 ,2 9 ) and E i (x,y)=23, then ESB i (x,y)=3; if V i (x,y)∈[2 9 ,10 3 ) and E i (x,y)=24, then ESB i (x,y)=3; if V i (x,y)∈[10 3 ,2 10 ) and E i (x,y)=24, then ESB i (x,y)=4; if V i (x,y)∈[2 10 ,2 11 ) and E i (x,y)=25, then ESB i (x,y)=4; if V i (x,y)∈[2 11 ,2 12 ) and E i (x,y)=26, then ESB i (x,y)=3; if V i (x,y)∈[2 12 ,2 13 ) and E i (x,y)=27, then ESB i (x,y)=3; if V i (x,y)∈[2 13 ,10 4 ) and E i (x,y)=28, then ESB i (x,y)=3; if V i (x,y)∈[10 4 ,2 14 ) and E i (x,y)=28, then ESB i (x,y)=4; if V i (x,y)∈[2 14 ,2 15 ) and E i (x,y)=29, then ESB i (x,y)=4; if V i (x,y)∈[2 15 , 65504] and E i (x,y)=30, then ESB i (x,y)=4.

Compared with the prior art, the present invention has advantages as follows.

(1) The embedding rate and the image fidelity of the present invention are high. In the high dynamic range image in OpenEXR format, when the embedding rate of the secret information is 30 bpps, the high dynamic range image after being embedded with the secret information is high in the image fidelity and small in the image quality volatility, thereby better meeting the human visual characteristics.

(2) The method provided by the present invention adopts the secret information and the carrier image embedding position to make the double-layer encryption, so as to effectively enhance the security, and the encryption process is low in complexity and has good practicability.

(3) The method provided by the present invention utilizes the conversion relationship between the data in OpenEXR Format and corresponding floating point data to modify the suitable floating point effective bits for embedding the secret information, which has low complexity, fast operation speed, excellent performance and universal practicability.

FIG. 1 is a flow chart of embedding secret information of a method provided by the present invention.

FIG. 2 is a flow chart of extracting secret information of the method provided by the present invention.

FIG. 3 shows image quality values of seven high dynamic range images embedded with the secret information relative to original high dynamic range images, and image quality values of the low dynamic range images which are processed through tone mapping the high dynamic range images embedded with the secret information, relative to the low dynamic range images which are processed through tone mapping the original dynamic range images.

The present invention is further described in detail accompanying with the drawings as follows.

The present invention provides a high dynamic range image information hiding method, which comprises steps of embedding secret information and extracting the secret information.

FIG. 1 shows the flow chart of the step of embedding secret information which specifically comprises:

{circle around (1)}_1 recording an original high dynamic range image in OpenEXR format to be embedded with the secret information as I org , wherein: a height of the I org is R and a width thereof is C,

recording an original secret information sequence for being embedded into the original high dynamic range image as W, wherein: a value of every information of the W is an integer from 0 to 9, encrypting the W through encryption algorithm, obtaining an encrypted secret information sequence, recording the encrypted secret information sequence as W hide , and taking the encryption algorithm as a secret key Key 1 , wherein: a length of the W is equal to R×C×3;

wherein in the step {circle around (1)}_1 of this embodiment, the encryption algorithm is Arnold transform or chaotic transformation, the W is scrambled through Arnold transform or chaotic transformation;

{circle around (1)}_2 according to a numerical conversion formula in OpenEXR format, obtaining three channel values of every pixel in the I org , recording an i th channel value of a pixel whose coordinates are (x,y) in the I org as V i (x,y), wherein

V i ⁡ ( x , y ) = { ( - 1 ) S i ⁡ ( x , y ) ⁢ 2 - 14 ⁢ ( 0 + M i ⁡ ( x , y ) 1024 ) E i ⁡ ( x , y ) = 0 , M i ⁡ ( x , y ) > 0 ( - 1 ) S i ⁡ ( x , y ) ⁢ 2 E i ⁡ ( x , y ) - 15 ⁢ ( 1 + M i ⁡ ( x , y ) 1024 ) 1 ≤ E i ⁡ ( x , y ) ≤ 30 ,

here, 1≤x≤R, 1≤y≤C, i is equal to 1, 2 or 3, a first channel value represents R channel value, a second channel value represents G channel value, a third channel value represents B channel value, S i (x,y) represents one sign bit of the i th channel of the pixel whose coordinates are (x,y) in the I org , E i (x,y) represents 5-bit exponent of the i th channel of the pixel whose coordinates are (x,y) in the I org , M i (x,y) represents 10-bit mantissa of the i th channel of the pixel whose coordinates are (x,y) in the I org ;

{circle around (1)}_3 according to every channel value and corresponding 5-bit exponent of every pixel in the I org , determining an embedding significance bit of the information to be embedded in every channel value of every pixel in the I org , recording the embedding significance bit (ESB) of the information to be embedded in the i th channel value of the pixel whose coordinates are (x,y) in the I org as ESB i (x,y), and recording the embedding significance bit of the information to be embedded in every channel value of every pixel in the I org as a secret key Key 2 ;

wherein in the step {circle around (1)}_3 of this embodiment, if V i (x,y)∈[2 E i (x,y)−15 ,2 E i (x,y)−14 ) and E i (x,y)∈[0,19], then ESB i (x,y)=5; if V i (x,y)∈[2 5 ,2 6 ) and E i (x,y)=20, then ESB i (x,y)=4; if V i (x,y)∈[2 6 ,10 2 ) and E i (x,y)=21, then ESB i (x,y)=3; if V i (x,y)∈[10 2 ,2 7 ) and E i (x,y)=21, then ESB i (x,y)=4; if V i (x,y)∈[2 7 ,2 8 ) and E i (x,y)=22, then ESB i (x,y)=3; if V i (x,y)∈[2 8 ,2 9 ) and E i (x,y)=23, then ESB i (x,y)=3; if V i (x,y)∈[2 9 ,10 3 ) and E i (x,y)=24, then ESB i (x,y)=3; if V i (x,y)∈[10 3 , 2 10 ) and E i (x,y)=24, then ESB i (x,y)=4; if V i (x,y)∈[2 10 ,2 11 ) and E i (x,y)=25, then ESB i (x,y)=4; if V i (x,y)∈[2 11 ,2 12 ) and E i (x,y)=26, then ESB i (x,y)=3; if V i (x,y)∈[2 12 ,2 13 ) and E i (x,y)=27, then ESB i (x,y)=3; if V i (x,y)∈[2 13 ,10 4 ) and E i (x,y)=28, then ESB i (x,y)=3; if V i (x,y)∈[10 4 ,2 14 ) and E i (x,y)=28, then ESB i (x,y)=4; if V i (x,y)∈[2 14 ,2 15 ) and E i (x,y)=29, then ESB i (x,y)=4; if V i (x,y)[2 15 ,65504] and E i (x,y)=30, then ESB i (x,y)=4;

{circle around (1)}_4 embedding information into every channel value of every pixel in the I org , wherein: embedding information into the i th channel value V i (x,y) of the pixel whose coordinates are (x,y) in the I org comprises:

{circle around (1)}_4a recording a value of an ESB i (x,y) th embedding significance bit of the information to be embedded of the V i (x,y) as D ESB i (x,y), setting a j th information in the W hide as a current information to be embedded in the W hide , recording the j th information as D hide (j), wherein: D ESB i (x,y)∈[0,9], an initial value of the j is 1, 1≤j≤J, the J represents a length of the W, D hide (j)∈[0,9];

{circle around (1)}_4b judging whether the D ESB i (x,y) is equal to the D hide (j), wherein: if the D ESB i (x,y) is equal to the D hide (j), the V i (x,y) is kept unchanged, so that embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org is completed; if the D ESB i , (x,y) is not equal to the D hide (j), the V i (x,y) is amended to complete embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org , wherein: the amendment comprises:

(1) setting V i,new (x,y)=V i (x,y)+N×SL i (x,y), (2) after the step (1), according to the V i,new (x,y), determining 5-bit exponent corresponding to the V i,new (x,y) and recording the 5-bit exponent as E i,new (x,y), and then through a manner as same as the step {circle around (1)}_3, according to the V i,new (x,y) and the E i,new (x,y), determining an embedding significance bit of the information to be embedded of the V i,new (x,y) and recording the embedding significance bit as ESB i,new (x,y); and then judging whether the D ESB i,new (x,y) of the ESB i,new (x,y) th embedding significance bit of the information to be embedded of the V i,new (x,y) is equal to the D hide (j), wherein if the D ESB i,new (x,y) is equal to the D hide (j), V i (x,y)=V i,new (x,y) is set, so that embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org is completed; if the D ESB i,new (x,y) is not equal to the D hide (j), a step (3) is executed; (3) setting V i,new (x,y)=V i (x,y)−N×SL i (x,y); (4) after the step (3), according to the V i,new (x,y), determining 5-bit exponent corresponding to the V i,new (x,y) and recording the 5-bit exponent as E i,new (x,y); and then through a manner as same as the step {circle around (1)}_3, according to the V i,new (x,y) and the E i,new (x,y), determining an embedding significance bit of the information to be embedded of the V i,new (x,y) and recording the embedding significance bit as the ESB i,new (x,y); and then judging whether the D ESB i,new (x,y) of the ESB i,new (x,y) th embedding significance bit of the information to be embedded of the V i,new (x,y) is equal to the D hide (j), wherein if the D ESB i,new (x,y) is equal to the D hide (j), V i (x,y)=V i,new (x,y) is set, so that embedding information into the i th channel value of the pixel whose coordinates are (x,y) in the I org is completed; if the D ESB i,new (x,y) is not equal to the D hide (j), a step (5) is executed; and (5) setting N=N+1, and then returning to the step (1) to go on, wherein: the V i,new (x,y) is an introduced intermediate variable, N is an integer whose initial value is equal to 1, SL i (x,y) is a step length corresponding to the E i (x,y), and “=” in the SL i (x,y)=2 E i (x,y)−15 , V i (x,y)=V i,new (x,y) and N=N+1 is an assignment symbol; and

{circle around (1)}_5 according to the numerical conversion formula in OpenEXR format, storing every channel value embedded with the information of every pixel in the I org , and obtaining a high dynamic range image embedded with the secret information.

FIG. 2 shows a flow chart of the step of extracting the secret information which comprises:

{circle around (2)}_1 recording the high dynamic range image embedded with the secret information as I stego , wherein: a height of the I stego is R and a width thereof is C;

{circle around (2)}_2 according to the numerical conversion formula in OpenEXR format, obtaining three channel values embedded with the information of every pixel in the I stego , recording an i th channel value embedded with the information of a pixel whose coordinates are (x,y) in the I stego as V i ′(x,y), wherein:

V i ′ ⁡ ( x , y ) = { ( - 1 ) S i ′ ⁡ ( x , y ) ⁢ 2 - 14 ⁢ ( 0 + M i ′ ⁡ ( x , y ) 1024 ) E i ′ ⁡ ( x , y ) = 0 , M i ′ ⁡ ( x , y ) > 0 ( - 1 ) S i ′ ⁡ ( x , y ) ⁢ 2 E i ′ ⁡ ( x , y ) - 15 ⁢ ( 1 + M i ′ ⁡ ( x , y ) 1024 ) 1 ≤ E i ′ ⁡ ( x , y ) ≤ 30 ,

here, 1≤x≤R, 1≤y≤C, i is equal to 1, 2 or 3, a first channel value represents R channel value, a second channel value represents G channel value, a third channel value represents B channel value, S i ′(x,y) represents one sign bit of the i th channel of the pixel whose coordinates are (x,y) in the I stego , E i ′(x,y) represents 5-bit exponent of the i th channel of the pixel whose coordinates are (x,y) in the I stego , M i ′(x,y) represents 10-bit mantissa of the i th channel of the pixel whose coordinates are (x,y) in the I stego ;

{circle around (2)}_3 according to the secret key Key 2 , obtaining a position where the information is embedded of every channel value embedded with the information of every pixel in the I stego , and recording the position in the V i ′(x,y) as ESB i ′(x,y);

{circle around (2)}_4 extracting information from every channel value embedded with the information of every pixel in the I stego , wherein: extracting information from the V i ′(x,y) comprises: recording a value of an ESB i ′(x,y) th embedded position in the S i ′(x,y) as D ESB i ′ (x,y), and then taking the D ESB i ′ (x,y) as the information extracted from the V i ′(x,y) and recording as D ex (u), wherein: D ESB i ′ (x,y)∈[0,9], an initial value of the u is 1, a step length thereof is 1, D ex (u)∈[0,9]; and

{circle around (2)}_5 forming sequences through the information extracted from all channel values embedded with information of all pixels in the I stego , recording the sequences as W ex , decrypting the W ex according to the secret key Key 1 , obtaining decrypted secret information sequences, and recording the obtained decrypted secret information sequences as W*.

In order to further illustrate the feasibility and effectiveness, the method provided by the present invention is tested.

In this example, a public high dynamic range image library Max Planck Institut informatik is selected, whose URL (Universal Resource Location) is resources.mpi-inf.mpg.de/hdr/gallery.html. There are seven high dynamic range images, whose specific information is shown in Table 1 as follows.

The secret information is embedded into the above seven high dynamic range images according to the step of embedding secret information of the method provided by the present invention, and then the secret information is extracted according to the step of extracting secret information of the method provided by the present invention; and then, through an existing HDR-VDP-2 method (reference: M. Narwaria, R. K. Mantiuk, M. P. D. Silva, P. L. Callet. “ HDR - VDP -2.2 : A Calibrated Method for Objective Quality Prediction of High Dynamic Range and Standard Images ”. In Journal of Electronic Imaging, 24(1), 2015), the image quality values of the high dynamic range images embedded with the secret information relative to the original high dynamic range images are evaluated; and through a PSNR method, the image quality values of the low dynamic range images which are processed through tone mapping the high dynamic range images embedded with the secret information, relative to the low dynamic range images which are processed through tone mapping the original dynamic range images, are evaluated, that is, the image fidelity is evaluated after the secret information is hidden; the image quality values obtained through the HDR-VDP-2 method are recorded as Q vdp , here, Q vdp ∈[0,100], the higher the value of the Q vdp , the higher the image fidelity; the image quality values obtained through the PSNR method are recorded as PSNR.

FIG. 3 shows values of the Q vdp and the PSNR corresponding to the seven high dynamic range images. It can be seen from FIG. 3 that when the embedded rate of the secret information is 30 bpps, Q vdp >91.61, PSNR>45.67, the image fidelity is higher. In FIG. 3 , Image represents the high dynamic range images.

The embedding significance bits embedded with information corresponding to the same 5 index bits of the channels of the pixels of the high dynamic range images are optional. Therefore, in practical application, the embedding significance bits and related data of the embedded information used in the method provided by the invention are listed in Table 2. MPWR i (x,y) of MPWR (Max Pixel-level Weber Ratio) of the i th channel value of the pixel whose coordinates are (x,y) in the I org of Table 2 are able to be obtained through

MPWR i ⁡ ( x , y ) = ⁢ T × SL i ⁡ ( x , y ) / PV i , min = ⁢ T × 2 E i ⁡ ( x , y ) - 25 / 2 E i ⁡ ( x , y ) - 15 = ⁢ 2 - 10 × T ,

wherein: T represents maximum number of iterations, T is obtained through the E i (x,y) and the ESB i (x,y), PV i,min represents a minimum value of the value range of the V i (x,y) and is obtained through the E i (x,y). It can be known from Table 2 that the MPWR i (x,y) varies with different value ranges of the V i (x,y) and is less than 0.04, which means that the method provided by the present invention has high image fidelity and small image quality volatility.

Y. Cheng, C. Wang, “A novel approach to steganography in high dynamic range images”, IEEE Multimedia, 16(3), pp. 70-80, 2009; M. Li, N. Huang, C. Wang, “A data hiding scheme for HDR images”, Int. J. Innovative Comput. Inf. Control, 7(5), pp. 2021-2035, 2011; and Y. Lin, C. Wang, W. Chen, F. Lin, W. Lin, “A novel data hiding algorithm for high dynamic range images”, IEEE Trans. Multimedia, 19(1), pp. 196-211, 2017 are top algorithms of the existing high dynamic range image information hiding. In this case, the three top algorithms and the method provided by the present invention are compared in the embedding rate and the image fidelity, and the comparison data are listed in Table 3. It can be seen from Table 3 that the embedding rate, the image fidelity and the image quality volatility of the high dynamic range image embedded with the secret information of the method provided by the present invention are better than the three top algorithms. The method provided by the present invention shows excellent performance and universal practicality.

This embodiment only shows that the high dynamic range image is fully embedded, that is, every channel of every pixel in the high dynamic range image is embedded with one information, an amount of all embedded information is R×C×3. In the actual process, it may be not fully embedded, such as the information is only embedded into every pixel of one or two channels in the high dynamic range image, so that an amount of all embedded information is less than R×C×3.