JP6403784B2 - Surveillance camera system - Google Patents

Description

  The present invention relates to a surveillance camera system that extracts information on an object such as a person, a car (including a two-wheeled vehicle), or an animal based on an image captured by a surveillance camera using an image recognition technique, and in particular, a processing load due to image recognition. In addition, the present invention relates to a surveillance camera system capable of reducing image quality and enabling efficient image search.

In recent years, the amount of data for transmitting images is increasing day by day. Due to the influence of terrestrial digital broadcasting that was switched in 2011, the resolution of commonly used images has changed from the conventional analog resolution (720pix × 480pix) to the Full HD (High-Definition) resolution (1920pix × 1080pix). It has changed. There is also a trend for higher resolution.
In surveillance cameras used in various locations such as stations, airports, buildings, and rivers, CCD (Charge Coupled Device) and CMOS (Complementary Metal-Oxide Semiconductor) sensors have become more sophisticated, and the resolution has been reduced from the conventional analog resolution. Dramatically improved.
On the other hand, image recognition technology for detecting various kinds of information such as a person, a car (including a two-wheeled vehicle), and an animal from an image taken by a surveillance camera is also improved.

2. Description of the Related Art Conventionally, there has been known a person search system that allows a computer to search for a desired person from video (moving images) captured or recorded by a surveillance camera or the like using an image recognition technique or the like (for example, Patent Documents). 1 to 4). Such a technique for searching based on the characteristics of an image itself without relying on external information such as tagging is generally called CBIR (Content-Based Image Retrieval), and has begun to be used for searching for a person.
Japanese Patent Application Laid-Open No. 2004-228688 extracts a portion in which a person (face) appears from an image, extracts a color histogram or the like as a feature quantity for individually identifying the person, and the feature quantity is similar to that of a desired person A video search system and a person search method that presume that they are the same person are disclosed.

  When it is possible to determine whether or not they are the same person with a certain degree of accuracy, as in Patent Document 2, by linking the same person taken at different times with a plurality of different cameras, It has become possible to track the movement of a person in an area that is too large to compare, and various applications have been proposed. For example, Patent Document 3 discloses an image security system that searches a captured video, specifies a movement path of a person, and uses the movement path for security management. Moreover, patent document 4 is disclosing the safety management apparatus which determines whether the user's action in a station platform is peculiar.

JP 2009-027393 A Japanese Patent No. 4700477 JP 2012-068717 A JP 2010-205191 A

FIG. 8 is a diagram illustrating an example of a configuration of a conventional general surveillance camera system.
As shown in FIG. 8, the conventional surveillance camera system 100 includes a plurality of surveillance cameras 110-1, 110-2,... (Hereinafter referred to as these surveillance cameras 110-1, 110-2). ,... Are simply referred to as the surveillance camera 110), and based on the video transmitted from the surveillance camera 110, an object such as a person, a car (including a two-wheeled vehicle), an animal, etc. A control terminal 150 having image recognition means for extracting information, a display terminal 60 for displaying object information extracted by the control terminal 150, a monitoring camera 110, a control terminal 150, and a display terminal 60 are connected to each other via a network line 70. It is a connected configuration.

As described above, in the surveillance camera system 100 used in various locations, the video from the surveillance camera 110 is normally sent to the control terminal 150 via a network line 70 such as a wireless LAN or a wired LAN, an IP line, a coaxial cable, or the like. It is done. In the control terminal 150, the image recognition unit analyzes and detects various information about a predetermined object from the video sent from the monitoring camera 110, and the video of the display terminal 60 that is monitored by the administrator is displayed. Output to the monitor. The administrator performs monitoring work using the displayed information.
In such a conventional surveillance camera system 100, as the number of surveillance cameras 110 that transmit images increases, the processing load on the control terminal 150 that performs information analysis increases in proportion to the number. Further, if the processing load on the control terminal 150 increases, the processing time naturally increases, and in the monitoring camera system 100 that requires real-time performance, this processing delay is a problem.

  The present invention has been made in view of such a conventional situation, and each monitoring camera includes an image recognition unit, and image information at the control terminal is exchanged between the monitoring cameras. It is an object of the present invention to provide a surveillance camera system that can reduce the processing load related to recognition and can reduce the processing load and processing time of the entire system.

  In order to achieve the above object, a surveillance camera system according to the present invention comprises a surveillance camera that takes an image of a surveillance area, a display terminal that displays an image taken by the surveillance camera, and a control terminal that controls the surveillance camera; In the surveillance camera system, the surveillance camera recognizes an image based on an image taken by itself, extracts specific object information, and the specific object extracted by the image analysis means. And an information communication means for exchanging information with other surveillance cameras.

  In the monitoring camera system according to the present invention for achieving the above object, the monitoring camera transmits the predetermined object information extracted by the image analysis means to the control terminal, and the control terminal The predetermined object information transmitted from the monitoring camera is notified to the display terminal, and the display terminal displays the predetermined object information notified from the control terminal on a display unit.

  In the monitoring camera system according to the present invention for achieving the above object, the predetermined object information transmitted from the monitoring camera to another monitoring camera or the control terminal includes a camera ID and an object detection time. A feature amount, time information, and position information are included.

  According to the present invention, each monitoring camera includes an image recognition means, and the processing information relating to image recognition at the control terminal can be reduced by exchanging the image-recognized information between the monitoring cameras. The processing load and processing time can be reduced.

It is a block diagram which shows an example of a structure of the surveillance camera system which concerns on Embodiment 1 of this invention. It is a block diagram which shows an example of a structure of the video storage server 30 of the monitoring camera 10 shown in FIG. It is a block diagram which shows an example of a structure of the image analysis part 40 of the monitoring camera 10 shown in FIG. It is a block diagram which shows an example of a structure of the display terminal 60 shown in FIG. It is a block diagram which shows an example of a structure of the control terminal 50 shown in FIG. It is a time chart for demonstrating the specific person search process in the surveillance camera system which concerns on Embodiment 1 of this invention. It is a time chart for demonstrating the specific person search process at the time of sharing information between the surveillance cameras in the surveillance camera system which concerns on Embodiment 2 of this invention. It is a figure which shows an example of a structure of the conventional general surveillance camera system.

Embodiment 1 Hereinafter, a surveillance camera system according to Embodiment 1 of the present invention will be described.
The surveillance camera system 1 according to Embodiment 1 of the present invention recognizes object information between a unit that recognizes an image based on an image captured by the surveillance camera and extracts information on the object, and another surveillance camera. By means of image recognition, the surveillance camera is informed by image recognition from among images taken by itself, for example, a person or a specific part of the person's face, etc., instructed by the supervisor. By extracting and sharing the extracted information with other surveillance cameras, the processing load related to image recognition on the control terminal is reduced, and the surveillance cameras exchange object information with each other, thereby processing the entire system. Load and processing time can be reduced.

[Control Configuration of Monitoring Camera System] A control configuration of the monitoring camera system according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing an example of the configuration of a surveillance camera system according to Embodiment 1 of the present invention.
The surveillance camera system 1 according to the first embodiment of the present invention includes surveillance cameras 10-1 and 10-2 (hereinafter simply referred to as surveillance camera 10 when the surveillance cameras 10-1 and 10-2 are not distinguished), The control terminal 50, the display terminal 60, the monitoring camera 10, the control terminal 50, and the display terminal 60 are connected to each other via a network line 70.
In FIG. 1, the same parts as those in FIG.

[Control Configuration of Monitoring Camera] As shown in FIG. 1, the monitoring camera 10 includes an imaging unit 20, a video storage server 30, and an image analysis unit 40.
The imaging unit 20 captures an image of the monitored area and outputs the captured image data to the video storage server 30 and the image analysis unit 40.

Next, the configuration of the video storage server 30 of the surveillance camera 10 according to the first embodiment of the present invention will be described with reference to FIG. FIG. 2 is a block diagram showing an example of the configuration of the video storage server 30 of the monitoring camera 10 shown in FIG.
The video storage server 30 sends a video transmission request to the imaging unit 20 and receives and stores image data from the imaging unit 20. The accumulated image is provided in response to a request from the image analysis unit 40 or the like.
As shown in FIG. 2, the video storage server 30 includes a video input unit 31, a recording / distribution control unit 32, a Web server unit 33, a storage 34, and a setting holding unit 35 as functional configurations.

  The video input unit 31 performs authentication for confirming the authority to acquire an image with the imaging unit 20, a video transmission request including designation of image quality, and acquisition of image data.

  The record distribution control unit 32 has a large-capacity cache memory, and manages writing and reading of image data to and from the storage 34. At the time of recording, in addition to the image data, an image ID (image identification information) serving as information for reading the image data is also recorded. The recording / distribution control unit 32 optimizes the recording arrangement in the image data writing unit and the storage 34 because the stream of the image data from the imaging unit 20 needs to be recorded in real time without being lost and meet the read request. In addition, scheduling of writing and reading is performed. In addition, in order to make the storage 34 RAID (Redundant Arrays of Inexpensive Disks), redundant data (parity) generation and writing control are performed. By dividing a plurality of image data streams into a plurality of write streams and generating horizontal parity or the like therefrom, reading for parity generation is not required.

  The Web server unit 33 receives an image request from the image analysis unit 40, and returns image data read out by requesting the recording / distribution control unit 32 as an image request response. Further, the operating state of the video storage server 30 is provided as a Web page.

  The storage 34 is a disk array composed of a plurality of HDDs (Hard Disk Drives) and SSDs (Solid State Drives), and uses a special file system suitable for simultaneous multiple access of video streams. For example, if all the correspondences between the image IDs that can uniquely identify one image and the recording positions are tabulated, it becomes larger than the capacity of the main memory, so that the correspondence can be referred to efficiently. . Data to be recorded is encrypted by a RAID controller or by each drive according to AES (Advanced Encryption Standard) -256 or the like.

  The setting holding unit 35 stores a schedule for the video input unit 31 to acquire a video from the imaging unit 20, a schedule for redistribution to the display terminal 60, and the like.

Next, the configuration of the image analysis unit 40 of the monitoring camera 10 according to the first embodiment of the present invention will be described with reference to FIG. FIG. 3 is a block diagram illustrating an example of the configuration of the image analysis unit 40 of the monitoring camera 10 illustrated in FIG. 1.
As shown in FIG. 3, the image analysis unit 40 includes an image acquisition unit 41, a face detection / feature amount calculation unit 42, a face registration / search unit 43, a face feature DB 44, a Web service unit 45, a search trigger as functional configurations. Unit 46, setting holding unit 47, and failure notification unit 48.
The image acquisition unit 41 acquires the video input from the imaging unit 20 and the recorded image data of the video storage server 30 by making an image acquisition request or the like.

  The face detection / feature amount calculation unit 42 divides the image acquired by the image acquisition unit 41 for each frame, and extracts an area estimated to be a human face from each frame. The estimation is basically performed based on the color of the skin and whether or not the eyes and nose can be detected, and for each extracted face area, the face direction is further estimated from the center of gravity of the face area and the relative position of the eyes and nose and mouth. The face area is handled as a rectangular image having a predetermined aspect ratio, but portions other than the face (background area) such as the four corners are filled with a prescribed color. Next, normalization processing such as size (number of vertical and horizontal pixels), resolution, brightness, and contrast (histogram) is performed on each face area. The size is normalized to a plurality of sizes. Next, the normalized face area is divided into blocks of a certain size for each of a plurality of sizes, and a color or luminance, gradient or edge thereof, or a histogram of gradient or edge pattern is obtained for each block, and the result Is output as a facial feature value. At that time, the photographing time, camera ID, image ID, and other metadata and face orientation information that the frame from which the extraction is performed have are output together.

  As metadata, the enlargement ratio at the time of size normalization, the accuracy of face estimation, the position in the frame from which the face is extracted, the extraction ID indicating the number of the plurality of faces extracted in the same frame, the event May be included. The accuracy of face estimation is a value that approaches 0 when another object covers a part of the face. The event type is information indicating whether the acquired image is based on normal recording (distribution) or alarm recording (distribution). Note that since the face feature amount has several thousand dimensions as it is, redundant components having high correlation with other components are removed. Further, a vector quantization method such as the Linde-Buzo-Gray method may be used in order to reduce the dimension of the feature amount.

  The face registration / search unit 43 writes (registers) the feature amount calculated by the face detection / feature amount calculation unit 42 into the face feature amount DB 44, and performs reading, searching, and other operations. The feature amount is registered as a record including attribute information such as a shooting time, a camera ID, a face image ID, a face orientation, and an enlargement ratio. At the time of registration, the face registration / retrieval unit 43 classifies the feature quantity into one of the categories and registers it in a cluster (leaf cluster) corresponding to the category. This clustering is performed in a multilevel manner using a dictionary (also called a map, a code book, etc.), a hash function, a covariance matrix, or the like. The data structure by this hierarchical clustering is the same as a multi-branch decision tree in the field of data mining, and is also called a tree in this example.

  As a method using a dictionary, vector quantization, self-organizing mapping, and the like are known. In the dictionary, multiple pairs of representative vectors (also called weight vectors) and the address of the dictionary or terminal cluster (leaf cluster) to be referred to next are recorded. The representative vector closest to the distance is found from all the dictionaries by a full search, and the process moves to the next dictionary, and finally the leaf cluster is reached. In the dictionary indicating the address of the leaf cluster, in addition to the address, the person ID of the person included in the leaf cluster is held. The leaf cluster stores data of records having a plurality of similar features. Each time a representative vector with the closest distance is found, the dictionary is updated by an algorithm called LBG or LVQ that slightly modifies the representative vectors of the cluster and neighboring clusters. In addition, it can also implement | achieve by the method of replacing with a dictionary and using as an address of the cluster what encoded the feature-value vector with hash functions (identifier), such as Locality Sensitive Hashing.

  A method using a covariance matrix or the like is based on a mixed distribution model, and performs a posteriori probability (logarithmic likelihood or Mahalanobis distance of a Mahalanobis distance) by performing a matrix operation between a feature quantity vector to be registered and an average value vector or a covariance matrix. Classification is performed by selecting the one having the largest sign (negative value), that is, the next lower cluster to be referred to is determined. An EM algorithm is known as an algorithm for optimizing parameters such as a covariance matrix. The k-d tree method that has been known for a long time is not suitable for searching for a high-dimensional vector as in this example. However, if only the components normalized and extracted to a small size among the components of the feature vector are used, there is a possibility that they can be used in the upper part of the tree.

  Each leaf cluster preferably has a size including exactly one distribution of a person's face image so as to correspond to the prototype. The person IDs held in the leaf cluster dictionary and the like are maintained in correspondence with the generation, division, and integration of the leaf clusters by the clustering algorithm described above. However, the person ID is not necessarily one-to-one corresponding to an actual individual person, and may be handled as a leaf cluster ID.

  In order to speed up the reference using the person ID as a key, a person ID table storing the address of the leaf cluster corresponding to the person ID may be provided. When reference is made using the face image ID as a key, a face image ID table is separately provided or a representative face image ID column is added to the person ID table. In any case, not all the faces that can be detected from the video of the video storage server 30 are registered in the face feature DB 44, so the search based on the face image ID is not complete. In other words, when it is desired to search using a face reflected in an arbitrary frame as a key, a feature amount is extracted from that frame (again).

  The feature amount search method is similar to registration, and the face registration / search unit 43 performs nearest neighbor search (NNS) sequentially from the highest cluster using the given feature amount as a key. In the leaf cluster, a linear (brute force) search is performed to obtain the Manhattan distance between the feature quantity and the key feature quantity of each record, and a desired similarity or a desired number of records is obtained. Note that if a cluster closest to the second, not the nearest neighbor, is selected only once in any of a plurality of classification hierarchies up to the leaf cluster, a maximum of several leaf clusters in the hierarchy is reached. If a linear search is performed among these, search omissions can be reduced. If the size of the leaf cluster is smaller than the distribution of the feature amount of one person, the search with the leaf cluster is unnecessary.

  The face feature DB 44 is a recording medium such as an HDD that holds record data included in the end cluster while maintaining information on a multi-level clustering structure (tree). In addition to storing parameters such as dictionaries, hash functions, and covariance matrices, and numerical values used by algorithms that optimize those parameters (such as values that are being calculated), a person ID table, a final search date / time list, and a black list are used as appropriate. And so on. These tree information, parameters, numerical values, and lists are all cached in the memory during operation. The end cluster is arranged at a recording position (sector) that can be read continuously. As a result, at least during a search, only one seek occurs for one search. The record of the face feature DB 44 is basically only new registration. The face feature DB 44 is internally composed of a plurality of divided DBs (different from the clustering described above). When the capacity of a certain divided DB reaches the upper limit, the oldest divided DB is initialized and Will be operated by a new registration destination. A time series of divided DBs is called a generation. Further, the image may be further divided according to the enlargement ratio, face orientation, and the like.

  The Web service unit 45 receives a search request from the display terminal 60, an image request, or setting information of the image analysis unit 40 of the monitoring camera 10 from the control terminal 50, and the processing result for the received request as data such as a Web page. Embody a search application by responding. The search request includes a specific person detection request (short-term search request), a key designation search request, a blacklist collation request, an event search request, and a combination thereof. When these are received, information on search conditions included in them (Face image ID or the like corresponding to the feature amount as a key) is transferred to the face registration / search unit 43. Based on the search result received from the face registration / search unit 43, Web page (display screen) data is generated and used as a response to the display terminal 60. The search result is provided as a tile display or the like in which a plurality of face images are arranged in a grid pattern. Each face image describes an address (URI) in the video storage server 30 based on the face image ID in the Web page. By making the display terminal 60 obtain and describing the face image ID itself and the person ID together, the display terminal 60 can make a new search request using the face in the search result as a key.

  The search trigger unit 46 controls the face registration / search unit 43 to automatically detect a specific person at an appropriate time interval. The search trigger unit 46 checks the current date and time and gives an appropriate search condition to the face registration / search unit 43.

  The setting holding unit 47 holds various settings necessary for the image analysis unit 40, information on users who can log in, and the like. The failure notification unit 48 notifies the control terminal 50 of a failure that has occurred in the image analysis unit 40 using an SNMP (Simple Network Management Protocol) trap or the like.

[Control Configuration of Display Terminal] Next, the configuration of the display terminal 60 of the surveillance camera system according to Embodiment 1 of the present invention will be described with reference to FIG. FIG. 4 is a block diagram showing an example of the configuration of the display terminal 60 shown in FIG.
As shown in FIG. 4, the display terminal 60 is a PC (personal computer) having a Web browser function, and transmits a search request and an image request to the image analysis unit 40 and the video storage server 30 of the monitoring camera 10. The Web page received as a response is displayed. The functions of the search request control unit 62, the search result display processing unit 63, the search condition designation processing unit 64, and the server status display processing unit 65 are javascript (trademark), Active X (trademark), and .Net included in the web browser and web page. The video display processing unit 61 is realized by Directshow (trademark) or the like. The display terminal 60 may have the same function as the display terminal of a general surveillance camera system. That is, an arbitrary surveillance camera 10 can be designated, and live video or recorded video can be acquired and displayed from the video storage server 30 by a protocol such as MRCP (Media Resource Control Protocol).

[Control Configuration of Control Terminal] Next, the configuration of the control terminal 50 of the surveillance camera system according to Embodiment 1 of the present invention will be described with reference to FIG. FIG. 5 is a block diagram showing an example of the configuration of the control terminal 50 shown in FIG.
The control terminal 50 is a general PC (personal computer) in order to cause the monitoring camera system of this example to capture an image of an external medium or to back up a recorded image. The external medium I / F 51 is an interface capable of connecting an arbitrary external medium such as a DVD drive or USB (trademark). The format conversion unit 52 converts the video and DB captured from the external medium I / F 51 and the like so as to match the format of the surveillance camera system.

The file unit 53 is a storage device that holds video captured from the external medium I / F 51 or the like or after format conversion. The upload unit 54 transmits a file by the FTP protocol or the like via the network line 70 between the image analysis unit 40 or the video storage server 30 of the monitoring camera 10 and the file unit 53. For example, the feature quantity and threshold value related to a specific person acquired from another monitoring camera 10 are uploaded to the image analysis unit 40 and merged with the face feature quantity DB 44 of the image analysis unit 40, or a snapshot of the face feature quantity DB 44 is taken as a file part. Or download to 53.
The external DB I / F 55 is an ODBC, JET database engine, or the like, and can access any server on the network and input / output data. For example, it is used when a feature amount and a threshold value related to a specific person of a similar surveillance camera system installed in another area are linked to each other.

  The network line 70 is a private network constructed by, for example, Ethernet (registered trademark) or the like, and connects each device from the video storage server 30 to the control terminal 50 of the monitoring camera 10 that can be installed at various bases. The network line 70 is not limited to a single collision domain network. The image analysis unit 40 or the like of the monitoring camera 10 can be constructed using a cloud computing infrastructure such as PaaS or IaaS.

[Specific Person Search Processing in Surveillance Camera System] Next, specific person search processing in the monitoring camera system according to Embodiment 1 of the present invention will be described with reference to FIG. FIG. 6 is a time chart for explaining a specific person search process in the surveillance camera system according to the first embodiment of the present invention.
The monitoring cameras 10-1 and 10-2 of the monitoring camera system according to the first embodiment of the present invention each include an image analysis unit 40 that identifies a person using an image recognition function. As a method for identifying a person, an existing technique disclosed in Japanese Patent Laid-Open No. 2013-242719 may be used.
At this time, the monitoring cameras 10-1 and 10-2 have a feature amount (additional information such as face information / clothing / walking speed) related to the person extracted by using the image recognition function and a camera ID / position for identifying the monitoring camera 10. Information / time information is communicated between the monitoring cameras 10 to share the information. The communication method between the monitoring cameras 10 is performed using IP packet communication or a dedicated line (coaxial / serial, etc.).

As shown in FIG. 6, in step S101, the control terminal 50 transmits a feature amount related to a specific person and a threshold value (matching rate) of the feature amount to the monitoring camera 10-1.
In step S <b> 102, the monitoring camera 10-1 transmits the feature amount related to the specific person and the threshold value of the feature amount transmitted from the control terminal 50 to the other monitoring camera 10 such as the monitoring camera 10-2. .

In step S <b> 103, each surveillance camera 10 analyzes a person who matches the received information related to the specific person by image recognition.
In step S104, when the monitoring camera 10-1 detects a person exceeding the threshold value transmitted from the control terminal 50 as a result of the person detection by the monitoring cameras 10-1 and 10-2, the monitoring camera 10-1 Then, the image analysis unit 40 temporarily accumulates as specific person information in the face feature value DB 44 (step S105), and for the control terminal 50 and the monitoring camera 10-2, the camera ID / feature value at the time of various detections / time information. / The position information is transmitted (step S106). << Here, it is assumed that a specific person is detected by the monitoring camera 10-1. >>

In step S107, the control terminal 50 that has received the specific person information from the monitoring camera 10-1 notifies the detection result to the display terminal 60 monitored by the administrator. On the other hand, the monitoring camera 10-2 that has received the specific person information from the monitoring camera 10-1 registers the detection result in the face feature DB 44 of the image analysis unit 40 (step S108).
In step S109, the display terminal 60 displays the specific person information notified from the monitoring camera 10-1 on a display unit (not shown).
In step S110, in response to an operation instruction from the administrator, the display terminal 60 controls the monitoring camera (here, the monitoring camera 10-2) that is in close proximity based on the position information / time information from the monitoring camera 10-1. And perform processing as an administrator, such as tracking a specific person.

  As described above, according to the surveillance camera system according to the first embodiment of the present invention, each surveillance camera includes an image recognition unit, and the image recognition information is mutually exchanged between the surveillance cameras. The processing load related to image recognition in the system can be reduced, and the processing load and processing time in the entire system can be reduced.

<Embodiment 2> Hereinafter, a surveillance camera system according to Embodiment 2 of the present invention will be described.
The surveillance camera system according to Embodiment 2 of the present invention has the same control configuration as that of the surveillance camera system according to Embodiment 1 of the present invention, but differs in the person specifying process described below.

[Specific Person Search Processing in Surveillance Camera System] Next, specific person search processing in the monitoring camera system according to Embodiment 2 of the present invention will be described with reference to FIG. FIG. 7 is a time chart for explaining a specific person search process when information is shared between monitoring cameras in the monitoring camera system according to the second embodiment of the present invention.

As shown in FIG. 7, in step S201, the control terminal 50 transmits a feature amount related to a specific person and a threshold value (matching rate) of the feature amount to the monitoring camera 10-1.
In step S202, the monitoring camera 10-1 transmits to the other monitoring camera 10 such as the monitoring camera 10-2 the feature amount related to the specific person and the threshold value of the feature amount transmitted from the control terminal 50. .

In step S <b> 203, each surveillance camera 10 analyzes a person who matches the received information on the specific person by image recognition with respect to the image taken by the surveillance camera 10.
In step S204, when the monitoring camera 10-1 detects a person exceeding the threshold value transmitted from the control terminal 50 as a result of performing the person detection with the monitoring cameras 10-1 and 10-2, the monitoring camera 10-1 Then, the image analysis unit 40 temporarily accumulates as specific person information in the face feature amount DB 44 (step S205), and for the control terminal 50 and the monitoring camera 10-2, the camera ID / feature amount / time information at the time of various detections. / The position information is transmitted (step S206). << Here, it is assumed that a specific person is detected by the monitoring camera 10-1. >>

In step S207, the control terminal 50 that has received the specific person information from the monitoring camera 10-1 notifies the detection result to the display terminal 60 monitored by the administrator. On the other hand, the monitoring camera 10-2 that has received the specific person information from the monitoring camera 10-1 registers the detection result in the face feature DB 44 of the image analysis unit 40 (step S208).
In step S209, the display terminal 60 displays the specific person information notified from the monitoring camera 10-1 on a display unit (not shown).
On the other hand, in step S210, the control terminal 50 that has received the specific person information from the monitoring camera 10-1 detects a monitoring camera other than the monitoring camera 10-1 that has detected the specific person (here, the monitoring camera 10-2). Then, it is determined whether or not to update the feature amount. Specifically, whether the data transmitted in step S201 (the feature amount related to the specific person and the threshold value of the feature amount) is retained, or the feature amount extracted by the monitoring camera 10-1 is used as future processing. Determine if.
Here, if the determination result is “YES”, the control terminal 50 advances the process to step S211 to transmit a feature amount update instruction to the monitoring camera 10-1, and if the determination result is “NO”. Then, the process ends.
In step S <b> 212, the monitoring camera 10-1 transmits the feature amount update instruction to the other monitoring camera 10-2, thereby sharing information with the other monitoring camera 10.
This assumes that the feature amount transmitted by the control terminal 50 in step S201 is not always accurate. In such a case, it is desirable to perform specific person detection after that using the feature quantity related to the specific person detected by the monitoring camera 10-1.

The monitoring cameras 10-1 and 10-2 that have received the new feature value from the control terminal 50 in steps S211 and S212 perform analysis of the specific person by image recognition using the new feature value (steps S213 and S214).
In step S215, when the monitoring camera 10-2 detects a person who has exceeded the new threshold value transmitted from the control terminal 50, in the monitoring camera 10-2, the image analysis unit 40 temporarily performs the facial feature value DB 44. (Step S216), and the camera ID / feature amount / time information / position information at the time of various detections are transmitted to the control terminal 50 and the monitoring camera 10-1 (step S217). << Here, it is assumed that a specific person is detected by the monitoring camera 10-2. >>

In step S218, the control terminal 50 that has received the specific person information from the monitoring camera 10-2 notifies the detection result to the display terminal 60 monitored by the administrator. On the other hand, the monitoring camera 10-1 that has received the specific person information from the monitoring camera 10-2 registers the detection result in the face feature DB 44 of the image analysis unit 40 (step S219).
In step S220, the display terminal 60 displays the specific person information notified from the monitoring camera 10-2 on a display unit (not shown).
In step S221, the display terminal 60 controls an adjacent monitoring camera (in this case, the monitoring camera 10-1) based on the position information / time information from the monitoring camera 10-2 in accordance with an operation instruction from the administrator. Process as an administrator, such as sending a signal and tracking a specific person.

  As described above, according to the surveillance camera system according to the second embodiment of the present invention, each surveillance camera includes the image recognition means, and the image recognition information is mutually exchanged between the surveillance cameras. The processing load related to image recognition in the system can be reduced, and the processing load and processing time in the entire system can be reduced.

<Embodiment 3> Hereinafter, a surveillance camera system according to Embodiment 3 of the present invention will be described.
The surveillance camera system according to Embodiment 3 of the present invention differs from the surveillance camera system according to Embodiment 1 of the present invention in the following contents. (1) When a surveillance camera having functions such as pan, tilt, and zoom is used in each surveillance camera 10, other surveillance cameras receive position information / time information from the surveillance camera that has detected the specified specific person information. In this case, the received surveillance camera may automatically control its orientation and the like in order to make it easier to detect the information specified from the position information.
Specifically, when the monitoring camera 10-1 detects specific person information and the monitoring camera 10-1 is photographing the entrance of the passage, the monitoring camera 10-2 photographs the vicinity of the exit of the passage. If it is, the monitoring camera 10-2 sharing the specific person information performs processing such as turning the direction near the passage exit and focusing.

  As described above, according to the surveillance camera system according to the third embodiment of the present invention, each surveillance camera includes an image recognition unit, and the image recognition information is mutually exchanged between the surveillance cameras. The processing load related to image recognition in the system can be reduced, and the processing load and processing time in the entire system can be reduced.

  It should be noted that the configurations and operations of the above-described embodiments are examples, and it is needless to say that they can be appropriately modified and executed without departing from the spirit of the present invention.

  The present invention relates to a surveillance camera system that extracts information on a moving object such as a person, a car (including a two-wheeled vehicle), or an animal based on an image captured by a surveillance camera using an image recognition technique. The present invention can be applied to a wide range of surveillance camera systems that can reduce the load and that enable efficient image retrieval.

1: surveillance camera system, 10, 10-1, 10-2: surveillance camera, 20: imaging unit, 30: video storage server, 31: video input unit, 32: record distribution control unit, 33: Web server unit, 34 : Storage, 35: Setting holding unit, 40: Image analysis unit, 41: Image acquisition unit, 42: Face detection / feature amount calculation unit, 43: Face registration / search unit, 44: Face feature amount DB, 45: Web service Unit: 46: search trigger unit, 47: setting holding unit, 48: failure notification unit, 50: control terminal, 51: external medium I / F, 52: format conversion unit, 53: file unit, 54: upload unit, 55 : External DB I / F, 60: Display terminal, 61: Video display processing unit, 62: Search request control unit, 63: Search result display processing unit, 64: Search condition designation processing unit, 65: Server status display processing unit, 66: Setting holding unit , 70: network line, 100: surveillance camera system, 110, 110-1, 110-2: surveillance camera, 150: control terminal.

Claims (3)

In a surveillance camera system comprising a surveillance camera that captures an area within a surveillance area, a display terminal that displays an image captured by the surveillance camera, and a control terminal that controls the surveillance camera,
The surveillance camera is
Image analysis means for recognizing an image based on an image taken by itself and extracting specific object information;
Information communication means for exchanging the specific object information extracted by the image analysis means with other surveillance cameras,
The monitoring camera communicates the feature amount related to the specific object information extracted using the image recognition function and the camera ID / position information / time information for specifying the monitoring camera between the monitoring cameras, and shares the information. ,
The control terminal transmits a feature amount related to the specific object information and a matching rate of the feature amount to the monitoring camera, and the monitoring camera transmits the other monitoring camera from the control terminal. Transmitting the feature amount related to the specific object information and the matching rate information of the feature amount,
Each of the surveillance cameras analyzes a person who matches the specific object information received with respect to the image captured by the user by image recognition,
The control terminal that has received the specific object information from the monitoring camera notifies the display terminal that an administrator is monitoring the detection result,
The other monitoring camera that has received the specific object information from the monitoring camera registers the detection result in the face feature DB of the image analysis unit,
The display terminal displays the specific object information notified from the monitoring camera, and controls a control signal to the adjacent monitoring camera based on position information / time information from the monitoring camera according to an operation instruction from an administrator. And monitoring based on the specific object information. The surveillance camera system according to claim 1,
Whether the control terminal that has received the specific object information from the monitoring camera updates the feature amount for the other monitoring camera other than the monitoring camera that has detected the specific object information. Determine whether to hold the feature amount and the matching rate of the feature amount related to the transmitted specific object information, or to determine whether to use the feature amount extracted by the monitoring camera as future processing,
When the determination result is “YES”, the control terminal transmits an update instruction for the feature amount to the monitoring camera, and when the determination result is “NO”, the process ends.
The monitoring camera performs information sharing with the other monitoring camera by transmitting an update instruction for the feature amount to the other monitoring camera;
When the surveillance camera detects an object exceeding the new matching rate transmitted from the control terminal, the image analysis means stores the specific object information in the face feature DB, and the control terminal and A monitoring camera system, wherein a feature amount / camera ID / position information / time information at the time of various detections is transmitted to the other monitoring camera. The surveillance camera system according to claim 1 or 2,
When using the surveillance camera having pan, tilt and zoom functions, received when the other surveillance camera receives position information / time information from the surveillance camera that has detected the specified specific object information. The surveillance camera system characterized in that the orientation of the surveillance camera is automatically controlled in order to facilitate detection of information designated from position information.

Images