JP2007233495A - Distributed image processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To extract all pieces of necessary information, even in a case where cameras are distributed in a wide monitoring area, by image monitoring of the wide monitoring area. <P>SOLUTION: In this distributed image processor, images of a monitoring camera 101 are stored, image frames to be processed in detail thereof are indexed in real time in a first image processor 102, and only the image frames to be processed in detail of the stored images are processed in detail based on the indexes in a second image processor 105. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image processing apparatus used for image monitoring.

  There is an increasing need for surveillance with television images in the areas of security and traffic control. In addition, a large number of image processing apparatuses are integrated at the center for the purpose of widening the monitoring area and saving labor.

  A plurality of image processing devices are installed in image monitoring targets scattered in a wide area, color component data extracted by the image processing device is collected in a central monitoring center via a network, and the image monitoring targets are [Patent Document 1] discloses a wide area monitoring apparatus that grasps the movement of an object between the image monitoring objects by comparing and integrating color component data.

  On the other hand, [Patent Document 2] discloses that the disclosed distributed image processing apparatus collects images of a monitoring camera at a central monitoring center via a network and performs image processing.

Japanese Patent No. 3577795 JP-A-8-251573

In the wide area image monitoring system disclosed in [Patent Document 1], information is extracted from an image by image processing devices at monitoring sites scattered in a wide area, and information already extracted is combined at a later center. Therefore, all information necessary for the image processing apparatus at the monitoring site must be extracted. For this reason, real-time processing and high-speed image processing devices that extract information with high accuracy are required for systems that require real-time performance, such as monitoring systems. Therefore, there is a problem that it is difficult to install a processing apparatus having sufficient processing capacity.
In addition, the configuration disclosed in [Patent Document 2] collects images of monitoring sites scattered over a wide area at a central monitoring center via a network, but there is no intruder and does not require information extraction processing. Since the data is transmitted to the network, there is a problem of requiring a broadband and large-scale network infrastructure.

  In addition, when monitoring an image over a wide area, it is necessary to store image data in an image processing apparatus at a monitoring site or a center in a later stage. However, in a conventional system, an image storage device having a huge capacity is provided for this purpose. There was a need.

  A first problem of the present application is to realize image monitoring of a wide monitoring area and to extract all necessary information even when cameras are scattered in the wide monitoring area.

  The second problem of the present application is to enable a central monitoring center to perform wide-area monitoring without using a broadband and large-scale network infrastructure.

  A third problem of the present application is to realize that only necessary images are stored even when cameras are scattered in a wide monitoring area and image monitoring is performed on the wide monitoring area.

In order to achieve the first object described above, the present invention relates to an image storage unit that stores an image from a camera, a first image processing device that performs processing using the image, and an image from the image storage unit. In a distributed image processing device comprising a second image processing device for processing,
The metadata management table showing the characteristics of the image from the camera, and the first image processing device creates an index using the image from the camera and stores it in the metadata management table, and stores the second image The processing device performs feature extraction processing using the corresponding image data from the image holding unit based on the index held in the metadata management table, and adds the extracted feature data to the index. Is.

In order to achieve the second problem described above, the present invention provides an image storage unit that stores an image from a camera, a first image processing device that performs processing using the image, and an image storage unit. In a distributed image processing apparatus comprising a second image processing apparatus that processes an image via a network,
A metadata management table indicating characteristics of images from the camera, wherein the first image processing apparatus creates an index using the image from the camera, and the network stores the index in the metadata management table; And the second image processing device performs feature extraction processing using the corresponding image data from the image holding means via the network, and adds the extracted feature data to the index. It is a feature.

In order to achieve the second problem described above, the present invention provides an image storage unit that stores an image from a camera, a first image processing device that performs processing using the image, and an image storage unit. In a distributed image processing apparatus comprising a second image processing apparatus that processes an image via a network,
The first image processing apparatus, comprising: a storage management unit that manages an image stored by the image storage unit and outputs it via the network when requested, and a metadata management table that indicates characteristics of the image from the camera. Creates an index using the image from the camera, stores the index in the metadata management table via the network, and the storage management means is based on the index indicated by the metadata management table. The image storage means manages the images stored.

  According to the distributed image processing apparatus of the present invention, the first image processing apparatus that puts an index in real time (real time) into the image to be stored is arranged at the monitoring site, and the corresponding center from the monitoring site is arranged at the central center. By performing feature extraction processing using image data and adding the extracted feature data to the index, there is no need to process the image only at the monitoring site.

  In addition, according to the distributed image processing apparatus of the present invention, only the image frames selected in the first image processing are sent to the second image processing apparatus in the central monitoring center via the network, so that the load on the network can be reduced. There is.

  Further, according to the distributed image processing apparatus of the present invention, when monitoring an image over a wide area, it is necessary to hold the image data in the image processing apparatus at the monitoring site or the center in the subsequent stage, but only the necessary image data is required. Is stored in the monitoring site apparatus, which eliminates the need to provide an image storage apparatus having a huge capacity compared to the conventional apparatus.

[Example 1]
FIG. 1 is a diagram of a first embodiment of a distributed image processing apparatus according to the present invention, which comprises an on-site monitoring apparatus 1 and a monitoring center 2. The on-site monitoring device 1 is placed at a monitoring site and has a monitoring camera 101, a first image processing device 102, an image storage function 103, and a storage management function 104. The monitoring center 2 has a second image processing device 105, a meta It consists of a data management table 106 and a metadata management function 107. The on-site monitoring device 1 and the monitoring center 2 are geographically separated, and these devices are connected by a communication line 180. The image of the monitoring camera 101 is stored in the image storage function 103 and processed at the same time as the first image processing apparatus 102 to detect and measure occurrences, and is used as an index when the stored image of the image storage function 103 is read out. Write to the management table 106.

  In this example, the ID number that identifies the surveillance camera 101, the imaging time, and the storage location of the image storage function 103 for the image, the generated item data detected and measured by the first image processing apparatus are collectively stored as metadata. Call. That is, the first image processing apparatus 102 generates new metadata and writes it in the metadata management table 106 every time the occurrence item is detected and measured. Since the storage capacity of the image storage function 103 is finite, in principle, it is necessary to erase the old image in order and write a new image. However, according to the request of the storage management function 104, the old image can be stored for a certain period. Or new images can be erased immediately. The storage management function 104 defines the image storage period of the image storage function 103 according to the metadata of the metadata management table 106. The metadata management table 106 has a finite number of metadata. In principle, the metadata is erased in order from the old metadata, and new metadata is written, but the metadata is immediately stored in accordance with a command from the metadata management function 107. Erase or store for a long time. The metadata management function 107 issues a command to the second image processing apparatus 105 according to the metadata stored in the metadata management table 106. The second image processing apparatus 105 reads an image stored in a predetermined location of the image storage function 103 in accordance with an instruction from the metadata management function 107, performs a predetermined image measurement, and uses the result as the corresponding metadata in the metadata management table 106. Add to the appropriate part of. The operator can browse and manipulate the metadata stored in the metadata management table 106 via the user interface.

FIG. 2 shows an example of a metadata configuration stored in the metadata management table 106. The metadata management items include a camera ID, an imaging time, a storage location, a storage period, occurrence items, occurrence item measurement quality, necessity of image features, image features, and presence / absence of image features. FIG. 2 shows that metadata 1 to n are already managed as an example of metadata. Assuming that the image stored by the image storage function 103 is digital data, when the image of the monitoring camera 101 is stored in the image storage function 103, the camera ID, the imaging time, and the The storage location and the storage period are determined and stored in the image storage function 103 based on this. At the same time, when the first image processing apparatus 102 detects and measures an occurrence, new metadata is written into the metadata management table 106 by the first image processing apparatus 102. In the example of metadata shown in FIG. The camera ID, the imaging time, the storage location, the occurrence item, the predetermined storage period, and the occurrence item measurement quality are determined. Hereinafter, these six items of metadata are set as metadata online items, and other metadata items are set as metadata offline items.

  Each time new metadata is generated in the metadata management table 106, the metadata management function 107 determines the necessity of image characteristics of the offline item based on the occurrence of the online item and the occurrence measurement quality of the metadata. If it is determined as NO, the image feature necessity item of the offline item is input as “NO”. On the other hand, if it is determined that it is necessary, the image feature necessity item of the offline item is set as “necessary”, and the data is initialized as “none” for the presence or absence of the image feature of the offline item. By rewriting the storage period of the online item as “postponed”, the image frame is stored until the processing of the second image processing apparatus 105 is completed. The metadata management function 107 issues a command to start processing after informing the second image processing apparatus 105 of the storage location of the image to be processed and the image feature to be extracted.

  The second image processing apparatus 105 extracts a predetermined feature from a predetermined image according to the command of the metadata management function 107, writes an image feature for the metadata in the metadata management table 106, and sets an item of presence / absence of an image feature. Enter “Yes”. If the corresponding image feature cannot be extracted, the operation is performed to input “impossible”.

  The metadata management function 107 refers to the metadata management table 106 in which the item of presence / absence of the image feature is rewritten as “present” or “impossible” with reference to the metadata added by the second image processing apparatus 105. On the other hand, the storage period is “extended” or rewritten by specifying a predetermined storage period.

  Here, the operation of the image storage function 103 will be described. The image storage function 103 stores a moving image digitally, but each image frame has a predetermined storage period and is automatically deleted. However, the storage period of the image frame corresponding to the metadata registered in the metadata management table 106 is under the management of the storage management function 104, and is stored or deleted according to the storage period of the metadata designated by the storage management function 104. .

  FIG. 3 is a diagram showing an image of the stored image sequence, and the horizontal axis represents the time axis. The hatched partial image frame is stored in a storage period predetermined by the image storage function 103. On the other hand, a portion corresponding to an image frame defined by the metadata is stored according to a predetermined storage period of the metadata. As described above, the storage period specified by the metadata is also determined in advance. However, when the storage period item is “extended” as described above, the second image processing apparatus 105 has completed the processing. Then, the data is stored indefinitely until the storage period is written in the storage period item, or stored according to the rewritten predetermined period.

  FIG. 9 is a diagram of a processing flow of the storage management function 104. The storage management function 104 receives the update information of the metadata information in the metadata management table 106 (Step 1). Confirm the storage period item of each metadata (step 2). If the metadata storage item is within the designated storage period or “extended”, a protection instruction for the corresponding frame is output to the image storage function 103. On the other hand, the image storage function 103 deletes image frames that have passed a predetermined storage period, but suspends deletion of image frames for which a protection instruction is issued from the storage management function 104.

  The operator can browse and operate the metadata managed in the metadata management table 106 via the user interface.

[Effect of the invention according to the first embodiment]
With the configuration described above, the first image processing apparatus performs online real-time processing, and the second image processing apparatus can efficiently and more complicatedly perform measurement processing using the processing result of the first image processing apparatus. Have. In addition, since the first image processing apparatus has a simple configuration that performs simple processing of simple event detection, the robustness of the apparatus can be easily realized, and the first image processing apparatus can be placed at various monitoring sites.

  As a result, the first image processing apparatus only has a motion detection function for detecting the presence or absence of an intruding object, and the second image processing apparatus performs pattern matching for discriminating attributes of the intruding object such as a vehicle or a person. Complex processing such as processing can be performed. Therefore, when used for monitoring roads, buildings, etc., the area where the second image processing device is located, for example, a central center that supervises the whole country, and the first image processing device as each region, for example, for each building or for each municipality It is possible to record necessary image frame data while reducing the signal transmission amount of the communication line 180 as much as possible.

[Example 2]
FIG. 4 shows a second embodiment of the distributed image processing apparatus of the present invention.

  It consists of an on-site monitoring device 3 and a monitoring center 4. The on-site monitoring device 3 is installed at a monitoring site, and includes a monitoring camera 101, a vehicle number recognition device 402, an image storage function 103, and a storage management function 104. The monitoring center 4 has a vehicle shape determination device 405 and metadata management. It consists of a table 406 and a metadata management function 407. The on-site monitoring device 3 and the monitoring center 4 are geographically separated, and these devices are connected by a communication line 180. The image of the monitoring camera 101 is stored in the image storage function 103, and at the same time, the vehicle number recognizing device 402 identifies an image frame in which the front of the vehicle (hereinafter referred to as the vehicle head) is shown and recognizes the vehicle head number. The metadata management table 406 generates metadata with the number recognition result every time the vehicle number recognition device 402 recognizes a new vehicle number.

  FIG. 5 is a diagram illustrating an example of metadata management items stored in the metadata management table 406, and includes items classified into online items and offline items. The online item includes a camera ID for identifying the camera, a time when the vehicle head reaches the screen field of view, a storage location when the image frame is stored in the image storage function 103, and a predetermined storage period. It consists of the recognition number recognized by the vehicle number recognition device 402 and the recognition rate reliability.

The recognition rate reliability holds the result of evaluating the reliability of the recognition result of the vehicle number recognition device 402. For example, if the difference between the first and second candidate probabilities is larger than a predetermined value in character recognition during number recognition, the reliability of the recognition rate reliability is judged to be “high” and input. Alternatively, if the character on the plate can be detected from the vehicle head image, the recognition rate reliability is evaluated as “high”, and if the character on the plate is not detected from the vehicle head image, the recognition rate reliability is evaluated as “low”. , Enter it. On the other hand, the offline item is composed of a vehicle shape feature necessity item, a vehicle shape item, and a measurement end flag item. When newly generating metadata, the metadata management table 406 determines the contents of the necessity / unnecessary items of the vehicle shape feature according to the online item. For example, when the recognition rate reliability is “low”, “necessary” is written in the necessity item of the vehicle shape feature, and when the recognition rate reliability is “high”, the necessity item of the vehicle shape feature is written. Writes “No”. This is because when the metadata recognition number is not reliable, the vehicle shape is measured by off-line processing to supplement the reliability. Alternatively, the registered vehicle number database 408 is connected to the metadata management table 406, and when the recognition number matches the registered number of the registered vehicle number database 408, “No” is written in the necessity item of the vehicle shape feature. . This is because it is considered that it is not necessary to newly measure the vehicle shape when it is determined that the vehicle has already been registered by referring to the recognition number. When the necessity item of the vehicle shape feature is “necessary”, the measurement end flag item is input “incomplete” to indicate that the measurement of the vehicle shape is not completed, and the corresponding image of the image storage function 103 is input. The storage period of the metadata is “extended” so that is not deleted. As already shown in FIG. 3, when data is stored in the image storage function 103, the image frame corresponding to the metadata is under the management of the storage management function 104, and “extension” is included in the storage period of the metadata Is stored permanently or for a predetermined period.

  The metadata management function 407 browses the metadata management table 406 to indicate that the necessity item of the vehicle shape feature of the metadata is “necessary”, and when the measurement end flag item of the metadata is “incomplete”, the vehicle An image corresponding to the shape determination device 405 is shown to determine the vehicle shape. When the determination process ends, the vehicle shape determination device 405 records the result in the vehicle shape item of the metadata and completes the measurement end flag. When the metadata management function 407 detects that the measurement end flag of the metadata in the metadata management table 406 is complete, the metadata management function 407 writes a predetermined storage period in the storage period item of the metadata.

  The vehicle shape determination device 405 may perform various product-sum calculation filters for contour enhancement and add the amount of the horizontal edge on the front surface of the vehicle as a feature amount to the vehicle shape item of the metadata. In addition, the vehicle shape model pattern database 409 is connected to the vehicle shape determination device 405 so that the vehicle head image is normalized based on the character size of the plate and the size of the plate, and then the degree of correlation is measured by a pattern matching technique. A model number indicating the model pattern of the vehicle registered in the vehicle shape model pattern database 409 is written in the vehicle shape item of the metadata, assuming that the model pattern having the largest degree is the shape of the vehicle. When the model registered in the vehicle shape model pattern database 409 is a luminance distribution pattern such as a bitmap image, the vehicle image to be compared can also be evaluated as a luminance distribution pattern by correlation calculation processing such as normalized correlation. . In addition, when the model registered in the vehicle shape model pattern database 409 is data representing the contour of the vehicle, the vehicle image to be compared is subjected to a contour detection filter and then subjected to higher-order local autocorrelation calculation to perform correlation. It is possible to evaluate the degree.

[Effect of the invention according to the second embodiment]
As described above, in this embodiment, in addition to the effects of the above-described embodiments, the vehicle number recognition and the vehicle shape determination process are separated, so that the vehicle shape is determined according to the recognition rate reliability and the recognition result of the vehicle number. Since the determination process can be switched between the “execution” state and the “omitted” state, there is an effect that an efficient system without unnecessary processing can be achieved.

  Further, every time a vehicle passes through the screen of the surveillance camera while saving the image of the surveillance camera, metadata is generated in real time processing by the number recognition device. The vehicle using the metadata and the saved image Since it is configured to determine the shape, the vehicle shape determination process does not require real-time processing. Therefore, the configuration and capability of the system that performs the processing of the vehicle shape determination processing device has the effect of increasing the degree of design freedom. is there.

Example 3
FIG. 6 is a diagram of a third embodiment of the distributed image processing apparatus according to the present invention, which comprises an on-site monitoring device 5 and a monitoring center 6. The on-site monitoring device 5 is installed at a monitoring site, and includes a monitoring camera 101, an intrusion detection device 602, an image storage function 103, and a storage management function 104. The monitoring center 6 includes a face authentication device 605 and a metadata management table 606. And a metadata management function 607 and a person database 608. The on-site monitoring device 5 and the monitoring center 6 are geographically separated, and these devices are connected by a communication line 180. The image of the surveillance camera 101 is stored in the image storage function 103, and at the same time, the intrusion detection device 602 selects an image frame in which an intruder is shown. The metadata management table 606 generates metadata every time the intrusion detection device 602 detects a new intruder.

FIG. 7 is a diagram illustrating an example of metadata management items stored in the metadata management table 606, and includes items classified into online items and offline items. The online item includes a camera ID for identifying a camera, a time when an intruder is detected, a storage location when an image frame at the corresponding time is stored in the image storage function 103, and a person whose intruder is a person It consists of sex. The person possibility can be determined by quantifying the size of the moving region, the continuity, and the amount of movement. On the other hand, the offline item includes a face authentication necessity item, an authentication result item, and an authentication end flag item. When newly generating metadata, the metadata management table 606 determines the contents of the necessity / unnecessary items of the face authentication according to the online item. For example, when the possibility of person is low, “Necessary” is written in the item for necessity of face authentication, and when the possibility of person is high, “No” is written in the item for necessity of face authentication. If the face authentication necessity item is “required”, the authentication end flag item is set to “incomplete” to indicate that the face authentication process is not completed, so that the corresponding image of the image storage function 103 is not deleted. In addition, the storage period of the metadata is “extended”. As already shown in FIG. 3, when data is stored in the image storage function 103, the image frame corresponding to the metadata is under the management of the storage management function 104, and “extension” is included in the storage period of the metadata Is stored permanently or for a predetermined period.

  The metadata management function 607 browses the metadata management table 606, and when the face authentication necessity item of the metadata is “necessary” and the authentication end flag item of the metadata is “incomplete”, the face authentication The image corresponding to the device 605 is shown and the face authentication process is performed. When the face authentication apparatus 605 authentication ends, the result is recorded in the authentication result item of the metadata, and the authentication end flag is set to “complete”.

[Effect of the invention according to the third embodiment]
As described above, in this embodiment, in addition to the effects of the above-described embodiments, intruder detection processing and face authentication processing are separated by mediating an image storage function, so that intruder detection is not performed at the monitoring site. Since there is no need for real-time processing in face authentication processing by placing an image processing apparatus capable of processing only real-time processing, there is an effect of increasing the degree of design freedom with respect to the configuration and capability of the entire system.

Example 4
FIG. 8 is a diagram of a fourth embodiment of a distributed image processing apparatus according to the present invention, in which a site monitoring apparatus 7 and a monitoring center 8 are connected by a wide area network (WAN) 811. The on-site monitoring device 7 includes an image monitoring terminal 803 including a monitoring camera 801 and a terminal image processing device 802, another image monitoring terminal having the same configuration as the image monitoring terminal 803, an image storage function 806, and a storage management function 807. The monitoring center 8 controls the image output function 818, the metadata management table 814, the center image processing device 815, the metadata management table 814, and the center image processing device 815. The image output function 818, the metadata management table 814, and the center image processing device 815 are connected via a local area network (LAN) 812. The local area network (LAN) 809 and the local area network (LAN) 812 are connected via a wide area network (WAN) 811.

The image of the monitoring camera 801 of the image monitoring terminal 803 is stored in the image storage function 806 via the LAN 808 and processed by the terminal image processing device 802 to perform vehicle number recognition or intrusion detection in the monitoring screen in real time. And metadata corresponding to the metadata management table 814 is generated based on the result. Since other image monitoring terminals such as the image monitoring terminal 804 and the image monitoring terminal 805 also generate metadata, these metadata are also held in the metadata management table 814. The configuration of this metadata is the same as that of the first embodiment. Metadata from a plurality of image monitoring terminals can be distinguished by the camera ID described with reference to FIG. The image stored in the image storage function 806 is deleted after a predetermined storage period, but the storage management function 807 stores and deletes the stored image frame corresponding to the metadata stored in the metadata management table 814. Implemented below. The operation of the storage management function 807 is the same as that of the storage management function 104 of the first embodiment. The metadata management function 816 operates the center image processing device 815 in accordance with the necessity of the image feature of the metadata stored in the metadata management table 814, and from the corresponding image of the image storage function 806 through the LAN 809, WAN 811, and LAN 812. The image feature is measured and the result is added to the corresponding metadata in the metadata management table 814. With the above configuration and operation, the metadata management table 814 stores, for example, n pieces of metadata as shown in the embodiment of FIG. 2, and the occurrence item and the image feature are used as keywords other than the camera ID and the storage time. You can search.

When the metadata is searched by the search, the corresponding image is loaded into the image output function 818 through the LAN 809, the WAN 811, and the LAN 812 and can be confirmed by the operator 820. The operator 820 can access the metadata management table 814 via the user interface and perform a metadata keyword search.
[Effect of the invention according to the fourth embodiment]
As described above, in this embodiment, in addition to the effects of the above-described embodiment, the image of the image monitoring terminal is placed in the image storage function connected to the monitoring site LAN, and only the necessary image is transferred to the WAN as necessary. In order to add processing to the monitoring center LAN via the network, there is an effect of avoiding congestion in the wide area network.

  In addition, the terminal image processing means at the monitoring site is provided with an image processing function having only a detection function, and a detailed wide-area monitoring system can be constructed by performing detailed measurement processing at the center as necessary.

  ADVANTAGE OF THE INVENTION According to this invention, it can implement | achieve providing the distributed image processing apparatus which responds to the request | requirement which performs image monitoring over many points, such as a river, a port, a road, and a railway.

1 is a diagram of a first embodiment of a distributed image processing apparatus according to the present invention. FIG. 6 is a configuration diagram of metadata of the first embodiment and the fourth embodiment of the distributed image processing apparatus of the present invention. The figure explaining matching of an image frame and metadata. The figure of the 2nd Example of the distributed image processing apparatus of this invention. The block diagram of the metadata of 2nd Example of the distributed image processing apparatus of this invention. The figure of 3rd Example of the distributed image processing apparatus of this invention. The block diagram of the metadata of 3rd Example of the distributed image processing apparatus of this invention. The figure of 4th Example of the distributed image processing apparatus of this invention. The figure of the processing flow of the preservation | save management function of the distributed image processing apparatus of this invention.

Explanation of symbols

101, 801 ... surveillance camera, 102 ... first image processing device, 103, 806 ... image storage function, 104, 807 ... storage management function, 105 ... second image processing device, 106, 406,
606, 814 ... metadata management table, 107, 407, 607, 816 ... metadata management function, 402 ... number recognition device, 405 ... vehicle shape determination device, 408 ... registered vehicle number database, 409 ... vehicle shape model pattern database, 602 ... Intrusion detection device, 605 ... Face authentication device, 608 ... Person database, 802 ... End image processing device, 803, 804, 805 ... Image monitoring terminal, 809, 812 ... Local area network, 811 ... Wide area network, 815 ... Center image processing device, 818... Image output function.

Claims (12)

Image storage means for storing images from the camera;
A first image processing device that performs processing using the image;
In a distributed image processing device comprising a second image processing device for processing an image from the image storage means,
A metadata management table showing the characteristics of the image from the camera;
The first image processing device creates an index using the image from the camera and stores it in the metadata management table,
The second image processing device performs feature extraction processing using the corresponding image data from the image holding unit based on the index held in the metadata management table, and adds the extracted feature data to the index. A distributed image processing apparatus. The distributed image processing apparatus according to claim 1,
The first image processing device creates a vehicle number recognition processing result as an index,
The second image processing apparatus reads an image used for number recognition of the image storage unit based on the index, measures a vehicle shape, and adds the result of the vehicle shape measurement to the index. Distributed image processing device. The distributed image processing apparatus according to claim 1,
The first image processing device creates an intruder detection result as an index,
Distributed image processing, wherein the second image processing apparatus reads an image with an intruder in the image storage unit based on the index, performs face authentication processing, and adds the face authentication result to the index apparatus. The distributed image processing apparatus according to claim 1,
Connecting the image holding means and the first image processing apparatus via a local area network;
A distributed image processing apparatus comprising a metadata management table and a second image processing apparatus in another network connected to the local area network. The distributed image processing apparatus according to claim 4,
The first image processing device detects the vehicle and performs number recognition processing to obtain an index for image storage means,
A distributed image characterized in that a second image processing apparatus reads an image used for number recognition of the image storage means based on the index, measures a vehicle shape, and adds the result of the vehicle shape measurement to the index. Processing equipment. Image storage means for storing images from the camera;
A first image processing device that performs processing using the image;
In a distributed image processing apparatus comprising: a second image processing apparatus that processes an image from the image storage unit via a network;
A metadata management table showing the characteristics of the image from the camera,
The first image processing device creates an index using an image from the camera, and stores the index in the metadata management table via the network,
The second image processing apparatus performs a feature extraction process using corresponding image data from the image holding unit via the network, and adds the extracted feature data to the index. Image processing device. The distributed image processing apparatus according to claim 6,
The first image processing device creates a vehicle number recognition processing result as an index,
The second image processing apparatus reads an image used for number recognition of the image storage unit based on the index, measures a vehicle shape, and adds the result of the vehicle shape measurement to the index. Distributed image processing device. The distributed image processing apparatus according to claim 6,
The first image processing device creates an intruder detection result as an index,
Distributed image processing, wherein the second image processing apparatus reads an image with an intruder in the image storage unit based on the index, performs face authentication processing, and adds the face authentication result to the index apparatus. The distributed image processing apparatus according to claim 6.
Connecting the image holding means and the first image processing apparatus via a first local area network;
A distributed image processing apparatus, wherein the metadata management table and a second image processing apparatus are connected to a second local area network connected via the network. Image storage means for storing images from the camera;
A first image processing device that performs processing using the image;
In a distributed image processing apparatus comprising: a second image processing apparatus that processes an image from the image storage unit via a network;
Managing images stored by the image storage means, and storage management means for outputting via the network if requested;
A metadata management table showing the characteristics of the image from the camera,
The first image processing device creates an index using an image from the camera, and stores the index in the metadata management table via the network,
The distributed image processing apparatus, wherein the storage management unit manages an image stored by the image storage unit based on an index indicated by the metadata management table. The distributed image processing apparatus according to claim 10.
The second image processing device different from the first image processing device;
The distributed image processing apparatus, wherein the storage management unit manages an image stored by the image storage unit based on a processing result of the image from the camera by the second image processing apparatus. The distributed image processing apparatus according to claim 10.
The distributed image processing apparatus, wherein the storage management unit holds an image from the camera when no data exists in the index indicated by the metadata management table.

Images