Image identifying method and related monitoring camera and monitoring camera system

1. Field of the Invention

The present invention relates to an image identifying method and a related monitoring camera and a related monitoring camera system, and more particularly, to an image identifying method of determining whether a target object is a leaving object or a missing object and a related monitoring camera and a related monitoring camera system.

2. Description of the Prior Art

A conventional monitoring camera compares a current monitoring image with a previous background image for detecting pixel variation when an object identifying process is executed. If the pixel variation in the images is small, the object captured by the images does not have an obvious change, which means a monitoring region of the monitoring camera is in a normal state. If the pixel variation in the images is large, the object stayed in the monitoring region is abnormal, but the conventional monitoring camera cannot identify what kind of changes happened to the object. When the object is made by soft material, such as a flag, the monitoring camera can detect deformation of the flag due to wind force; when the object is made by rigid material, the monitoring camera can detect a movement of the object due to accident impact; when the object is in an outdoor place, the monitoring camera can detect intensity variation of the object influenced by the weather. The user only focuses whether the object is delivered or stolen, and does not care the foresaid conditions about the deformation, the movement or the intensity variation of the object. Therefore, the conventional monitoring camera wastes computation time to execute the object identifying process, but cannot provide accurate identifying result because of the foresaid conditions.

The present invention provides an image identifying method of determining whether a target object is a leaving object or a missing object and a related monitoring camera and a related monitoring camera system for solving above drawbacks.

According to the claimed invention, an image identifying method of determining whether a target object is a leaving object or a missing object includes acquiring a foreground region corresponding to the target object within a monitoring image, analyzing whether the target object inside the foreground region conforms to a variant feature, and comparing the foreground region with a reference image to determine whether the target object inside the foreground region belongs to the leaving object or the missing object when the foreground region does not conform to the variant feature.

According to the claimed invention, a monitoring camera includes an image receiver and an operation processor. The image receiver is adapted to receive a monitoring image. The operation processor is electrically connected to the image receiver. The operation processor is adapted to acquire a foreground region corresponding to the target object within a monitoring image, analyze whether the target object inside the foreground region conforms to a variant feature, and compare the foreground region with a reference image to determine whether the target object inside the foreground region belongs to the leaving object or the missing object when the foreground region does not conform to the variant feature.

According to the claimed invention, a monitoring camera system includes at least one monitoring camera and an operational processing device. The monitoring camera is adapted to receive a monitoring image. The operational processing device is connected to the monitoring camera in a wire manner or a wireless manner. The operational processing device is adapted to acquire a foreground region corresponding to the target object within a monitoring image, analyze whether the target object inside the foreground region conforms to a variant feature, and compare the foreground region with a reference image to determine whether the target object inside the foreground region belongs to the leaving object or the missing object when the foreground region does not conform to the variant feature.

The image identifying method and the related monitoring camera and the monitoring camera system of the present invention can track and record the passerby's path, and firstably obviate conditions of generating error alert due to the intensity variation, the movement or the deformation of the target object when no person approaches. Generally, the condition about the intensity variation can be analyzed and filtered before judgments about the movement and the deformation, which depends on the actual demand. After obviation of the error alert, the image identifying method can determine whether the target object belongs to the leaving object or the missing object, and send out a warning message related to determination about the leaving object or the missing object. For example, the monitoring camera or the monitoring camera system may be installed on an outdoor hallway, and the image identifying system can immediately identify a state of merchandise when the merchandise is delivered or stolen, and send the message to remind the user of the merchandise's condition accordingly. Thus, the present invention can effectively decrease the amount of data and computation in the image identifying process by excluding the object formed by the variant feature, so as to identify the target object belongs to the leaving object or the missing object rapidly and accurately.

These and other objectives of the present invention will no doubt become obvious to those of ordinary skill in the art after reading the following detailed description of the preferred embodiment that is illustrated in the various figures and drawings.

FIG. 1 is a functional block diagram of a monitoring camera according to an embodiment of the present invention.

FIG. 2 is a functional block diagram of a monitoring camera system according to the embodiment of the present invention.

FIG. 3 is a flow chart of the image identifying method according to the embodiment of the present invention.

FIG. 4 to FIG. 7 are diagrams of the monitoring image acquired at different points of time according to the embodiment of the present invention.

FIG. 8 and FIG. 9 are diagrams of position relation between a foreground region and a moving object drawn in different ways according to the embodiment of the present invention.

Please refer to FIG. 1 and FIG. 2 . FIG. 1 is a functional block diagram of a monitoring camera 10 according to an embodiment of the present invention. FIG. 2 is a functional block diagram of a monitoring camera system 12 according to the embodiment of the present invention. The monitoring camera 10 can include an image receiver 14 and an operation processor 16 electrically connected to each other. The image receiver 14 is used to capture or receive a monitoring image. The operation processor 16 can analyze the monitoring image to execute an image identifying method, for determining whether a leaving object or a missing object is within the monitoring image. Moreover, the monitoring camera 10 can be matched with an operational processing device 18 to establish the monitoring camera system 12 , and the operational processing device 18 can be defined as an independent controller. An amount of the monitoring camera 10 can be one or more. The monitoring camera 10 may not execute the image identifying method, but transmits the monitoring image to the operational processing device 18 ; the operational processing device 18 can execute the image identifying method to determine whether the leaving object or the missing object is within the monitoring image. The monitoring camera 10 and the operational processing device 18 can be connected to each other in a wire manner or in a wireless manner, which depends on actual demand.

Please refer to FIG. 1 to FIG. 7 . FIG. 3 is a flow chart of the image identifying method according to the embodiment of the present invention. FIG. 4 to FIG. 7 are diagrams of the monitoring image acquired at different points of time according to the embodiment of the present invention. In the image identifying method, step S 100 is firstably executed to acquire a foreground region R 1 , which has a position corresponding to the target object O, within a monitoring image I 3 . Step S 100 can compare the monitoring image I 3 with a background image I 1 , and determine a possible range where the target object O may appear via pixel variation between different images. The possible range can be represented as the foreground region R 1 . Then, steps S 102 and S 104 are executed to define a region of interest R 1 based on the foreground region R 1 , and to determine whether a specific object P 1 is in the region of interest R 1 during a first predefined period started at an acquiring point of time about the foreground region R 1 . The region of interest R 1 can be a frame or a circle around the foreground region R 1 ; for example, an interval between boundaries of the foreground region R 1 and the region of interest R 1 may be ranged from three meter to five meter. The first predefined period may be ranged from five seconds to ten seconds. The specific object P 1 inside the monitoring image I 4 can be a passerby who is used to determine whether the target object O is at rest.

If the specific object P 1 is in the region of interest R 1 during the first predefined period, the target object O may not be at rest. The specific object P 1 can be a deliveryman who puts down the target object O and is ready to leave, or someone who passes through or is ready to take away the target object O, then step S 106 is execute to pause or end step of determining whether the target object O belongs to the leaving object or the missing object. If the specific object P 1 is not in the region of interest R 1 during the first predefined period, the target object O may be at rest, and step S 108 is executed to analyze whether the target object O inside the foreground region R 1 conforms to a predetermined variant feature. The image identifying method can utilize the pixel variation between different images to determine whether the target object O is put into a designated area or taken away from the designated area. However, if position of the target object O is accidently or purposely changed, or a shape of the target object O is deformed, or a shadow of the target object O is varied due to ambient illumination, existence of the target object O may be easily misjudged, so that step S 108 can be used to obviate conditions of the target object O not belonging to the leaving object and the missing object.

Thus, intensity variation, deformation or a movement of the target object O can be interpreted as the variant feature in step S 108 . The image identifying method can utilize convolutional neural networks (CNN) classifying technology to determine whether the foreground region R 1 conforms to the variant feature; or else, the image identifying method may utilizes feature detection technology to determine whether the target object O inside the foreground region R 1 is moved. First, one or more first feature points can be acquired within the foreground region R 1 , and the background image I 1 can be defined as a reference image for searching one or more second feature points, which corresponds to the first feature points, inside a background region R 2 of the reference image. Position of the background region R 2 can be related to the foreground region R 1 . If movement differences between each first feature point and a related second feature point are entirely or mostly smaller than a predefined threshold, the movement difference of all feature points can be the same or similar to each other, so that diversity of the foreground region R 1 and the background region R 2 can represent that the target object O is moved.

Besides, the image identifying method may further compute first histogram distribution information of the foreground region R 1 and second histogram distribution information of the background region R 2 . Then, the image identifying method can set a specific range, which is similar to the first histogram distribution information, inside the second histogram distribution information, and adjust convergence position of the specific range for re-computing its histogram distribution information. If the local histogram distribution information of the specific range is different from the first histogram distribution information, the target object O inside the foreground region R 1 is irrelevant to the object inside the background region R 2 . If the local histogram distribution information is similar to the first histogram distribution information, and/or a dimension of the object inside the foreground region R 1 is equal to or similar to a dimension of the object inside the background region R 2 , difference between the foreground region R 1 and the background region R 2 can be resulted from the movement or the deformation of the target object O.

Before step S 108 , step S 107 can be optionally executed to detect a position relation between the foreground region R 1 and a moving object P 2 during a second predefined period back from the acquiring point of time. The second predefined period can be ranged from ten seconds to thirteen seconds earlier than the acquiring point of time. The moving object P 2 inside the monitoring image I 2 can be the passerby; the specific object P 1 and the moving object P 2 may represent the same person or two different persons, which depend on actual demand. If the passerby stays around the target object O during the second predefined period, variation in the target object O may be man-made, so that step of determining whether the target object O belongs to the leaving object or the missing object can be executed continuously. Therefore, step S 106 can be executed when the position relation does not conform to a predefined condition, and step S 108 can be executed when the position relation conforms to the predefined condition.

It should be mentioned that the predefined condition may represent the moving object P 2 is pretty close to the foreground region R 1 . Please refer to FIG. 8 and FIG. 9 . FIG. 8 and FIG. 9 are diagrams of the position relation between the foreground region R 1 and the moving object P 2 drawn in different ways according to the embodiment of the present invention. As shown in FIG. 8 , the predefined condition can represent that a minimal distance between the foreground region R 1 and a moving path Pth of the moving object P 2 is smaller than a predefined value Thd. As shown in FIG. 0.9 , the image identifying method can mark a rectangular frame around the object; if the moving object P 2 is pretty close to the target object O, a marking frame of the foreground region R 1 can be enlarged as the rectangular frame R 1 ′ drawn by a dotted-line, which means a dimension of the marking frame R 1 generated at the acquiring point of time can be smaller than a dimension of the marking frame R 1 ′ during the second predefined period.

In step S 108 , if the target object O inside the foreground region R 1 conforms to the variant feature, pixel variation in the foreground region R 1 and the background region R 2 can just be the intensity variation, the deformation or the movement of the target object O, so that the image identifying method can determine the target object O inside the foreground region R 1 does not belong to the leaving object or the missing object, and step S 106 is executed accordingly. If the target object O inside the foreground region R 1 does not conform to the variant feature, step S 110 can be executed to compare the foreground region R 1 with the reference image for determining whether the target object O inside the foreground region R 1 belongs to the leaving object or the missing object.

As an example of comparing the foreground region R 1 of the monitoring image I 3 with the background image I 1 , when the background image I 1 is used as the reference image, and the background region R 2 does not have the target object O but the foreground region R 1 has the target object O, the target object O can belong to the leaving object put by the moving object P 2 . As an example of comparing the foreground region R 1 of the monitoring image I 3 with the monitoring image I 4 , when the monitoring image I 4 is used as the reference image, and the foreground region R 1 has the target object O but the background region R 2 of the monitoring image I 4 does not have the target object O, the target object O can belong to the missing object taken away by the specific object P 1 .

The image identifying method and the related monitoring camera and the monitoring camera system of the present invention can track and record the passerby's path, and firstably obviate conditions of generating error alert due to the intensity variation, the movement or the deformation of the target object when no person approaches. Generally, the condition about the intensity variation can be analyzed and filtered before judgments about the movement and the deformation, which depends on the actual demand. After obviation of the error alert, the image identifying method can determine whether the target object belongs to the leaving object or the missing object, and send out a warning message related to determination about the leaving object or the missing object. For example, the monitoring camera or the monitoring camera system may be installed on an outdoor hallway, and the image identifying system can immediately identify a state of merchandise when the merchandise is delivered or stolen, and send the message to remind the user of the merchandise's condition accordingly. Thus, the present invention can effectively decrease the amount of data and computation in the image identifying process by excluding the object formed by the variant feature, so as to identify the target object belongs to the leaving object or the missing object rapidly and accurately.

Those skilled in the art will readily observe that numerous modifications and alterations of the device and method may be made while retaining the teachings of the invention. Accordingly, the above disclosure should be construed as limited only by the metes and bounds of the appended claims.