JP2005149345A - Automatic searching system and searching method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an automatic searching system and an automatic searching method. <P>SOLUTION: Wide-angle image signals within a range covered with a first image acquisition device are consecutively caught, image discrimination is performed by a processor connected to the database, the first image acquisition device, and a second image acquisition device, and thus a portion matching a first feature image in the database in a wide-angle image is set as an expansion target, and the second image acquisition device is driven to catch an expanded image of the expansion target, and in the case that it is found that the expanded image has a portion matching a second or last feature image recorded in the database on the basis of the caught expanded image, a searching object is regarded to be found to start a searching report means. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an automatic searching system and a searching method, particularly using an image identification technique, performing a stepwise comparison on a plurality of search-symmetric feature images, and The present invention relates to an automatic searching system and a searching method that improve resolution and discrimination effect by actually catching an enlarged image.

  Conventionally, as a method of searching for a specific target such as a criminal or a stolen vehicle, a method that is performed by matching each feature in a comparative reference material by relying on the human naked eye, Most of the features are various images. For example, when searching for a stolen vehicle, the feature material includes the vehicle type, vehicle color, tag number, etc., and the person who searches for the vehicle is more interested in comparing the vehicle color and vehicle style. Features. When the vehicle corresponding to the priority feature is found, the billboard number of the next step is compared and compared, and if the tag number matches, it is considered that the vehicle has been found and the vehicle owner is notified or sent to the management office Carry and carry out subsequent work such as storage.

  However, with the rapid development of the computer and information industries, it is quite inefficient and costly to perform the above-mentioned searching comparisons at a given point by personnel or on dispatched vehicles. It is a means of such deterioration. Personnel may also adversely affect identification accuracy due to factors such as fatigue and weak concentration. Also, there are often times when it is not possible to work full-time due to personnel being crushed or rested. Overall, conventional methods can degrade the efficiency of searching operations to a certain degree.

  In order to overcome the high cost and low efficiency issues of human comparison, in recent years, mature image identification techniques have been developed and used for some degree of identification applications.

  Image identification technology is already mature and widely applied, but all practical applications at the moment are only identification of fixed mobile phones at a given distance, or passive, which requires immediate input of comparative data manually It is a form. As a result, there is still no departure from manually searching for specific purposes by personnel, and it is necessary to place a number of differently functioning sensors to provide signals for various operations. If the cost is increased or damaged, it is necessary to provide a maintenance budget, which may greatly increase the overall burden.

  Each patent also needs to help identify by performing the magnification required by a single captured image, but is partially expanded despite the number of pixels in the captured image. In this case, it is necessary to improve the difficulty of identification and to perform an artificial comparison. Is difficult to achieve. For example, after shooting a crime scene suspect by monitoring recording and enlarging the face part, it is often impossible to confirm the details of the face due to the above-mentioned digital enlargement blurring phenomenon. There is considerable technical difficulty with the control.

  Also, in modern times, certain nations or certain units have image enhancement technology that can analyze or expand the required image, for example, by a dim comparison or other method. However, if not all countries or units have such image enhancement technology and severe cases arise, depending on the few countries or units mentioned above will provide cooperation and improve image resolution, but both documents The round-trip time may lose the early nature of elucidating crime cases, the general public simply has it, no help is required, and image enhancement technology is not perfect, limited Only improvements can be implemented.

  In view of the problem that the conventional means using the image identification technique does not have the effect of manually searching for a specific target, and the caught image cannot still be effectively identified after enlargement. The inventor uses the existing executable image identification technique and manually executes the stepwise comparison target by catching the enlarged images corresponding to the plurality of feature images corresponding to the search target. It is intended to be able to provide a manual searching technique that can overcome the shortcomings of the prior art and manually catch a clear image for identification.

  Therefore, the present invention is a manual search that can improve resolution and discrimination effect by performing stepwise comparison and contrast for a plurality of feature images as a search target and actually catching an enlarged image. The main purpose is to provide an operating system.

  The present invention also provides a manual search that can improve the resolution and the discrimination effect by executing a stepwise comparison and contrast on a plurality of feature images to be searched and actually catching an enlarged image. The other purpose is to provide a scanning method.

  To this end, the manual searching system according to the present invention comprises a database for recording at least first and second feature images to be searched, and a wide angle within the covered range. A first video acquisition device that continuously captures images, a second video acquisition device that dynamically captures enlarged images, and connected to the database and the first and second video acquisition devices to perform image identification At the same time, the first feature image matching part of the wide-angle image is set to be the enlargement target, the second image acquisition device is driven to catch the enlargement image of the enlargement target, and then the first image in the enlargement image When a matching part is found in the two-feature image, it is considered that the search target is being searched, and at the same time Having a processing unit for starting the searching reporting means, at least.

In order to achieve the other object, the manual searching method according to the present invention includes the following steps.
Providing a database recording at least first and second feature images to be searched;
Continuously catching wide-angle images within the range covered by the first video acquisition device; and b.
Setting a portion matching the first feature image of the wide-angle image to be an enlargement target by performing image identification by the processing device; and c.
A step d of catching an enlarged image of the enlargement target by a second video acquisition device;
Considering the search target to be found if the processing device finds that it has a portion matching the second feature image in the magnified image; and
And at least a step f of activating the searching reporting means.

The characteristic image of the search target is different according to the actual target, and the quantity is different. Therefore, when there are two or more feature images to be searched, in order to achieve the above-mentioned purpose, the present invention realizes a kind of automatic searching system, which includes a plurality of feature images to be searched. A first video acquisition device that continuously captures a wide-angle image within a covered range;
A second video acquisition device that dynamically captures an enlarged image; connected to the database and the first and second video acquisition devices; performs image identification; and drives the second video acquisition device first. Catch the enlarged image of the part that matches the first feature image in the wide-angle image, then catch the enlarged image of the part that matches the other feature image in succession, and one before matching the last feature image The operation is continued until a portion of the enlarged image is found, and when it is found, the processing target is regarded as having been already searched, and at the same time, has a processing device that activates the searching report means.

Similarly, when there are two or more feature images to be searched, in order to achieve the other object, the present invention realizes a kind of automatic searching method.
Providing a database for recording a plurality of feature images to be searched; and
Continuously catching wide-angle images within the range covered by the first video acquisition device; and b.
Setting a portion matching the first feature image in the wide angle image to be an enlargement target by performing image identification by the processing device;
Step d for catching an enlarged image of the enlargement target by the second video acquisition device;
Sequentially identify magnified images that match other feature images and perform catch, and consider finding the search target when the last feature image matches the previous magnified image e,
And a step f of activating the searching reporting means.

  In order to elucidate the object, features, and effects of the present invention, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.

  FIG. 1 is a block diagram illustrating the automatic searching system of the present invention. As shown in the figure, the automatic searching system according to the present invention includes at least a processing device 1, a database 2, a first video acquisition device 3, and a second video acquisition device 4.

  Among them, the database 2 is built in a hard disk, memory, or other recording medium of a computer, server, or other similar electronic device, and a first feature image, a second feature image, and a third feature to be searched. You can record images. In the present embodiment, an example of finding a stolen vehicle with the help of the police will be described. Therefore, for example, the style and color of the stolen vehicle is adopted as the first feature image, the contour of the vehicle tag is adopted as the second feature image, and the third feature image is used as the third feature image. You can adopt the tag number. Among them, in the present embodiment, the present invention can adopt, for example, the outline of a vehicle tag as searching in the database 2 and the number of the tag as a third feature image. Among them, in the present embodiment, the present invention will be described by taking as an example recording the first feature image, the second feature image, and the third feature image to be searched in the database 2. However, it should be noted that the database 2 can selectively record more feature images, and is not limited to simply recording three feature images, and the database 2 has more searches. The feature image to be updated can be updated or increased at any time.

  The first video acquisition device 3 is a means capable of continuously catching a wide angle image within a covered range, and a fixed wide angle photographing device can be selectively employed. Further, the second video acquisition device 4 is used for the purpose of dynamically catching and enlarging an image, and a movable telephoto photographing device can be selectively employed. It is possible to catch an enlarged image that is controlled and set as an enlargement target. Among them, since the fixed wide angle photographing apparatus is a conventional one, the detailed explanation is omitted here. In addition, in the present embodiment, image catching is performed by one first video acquisition device 3 and one second video acquisition device 4, but it is understood that the first video acquisition device 3 The arrangement quantity of the second video acquisition device 4 is not limited to them. For example, by arranging a plurality of second video acquisition devices 4, a plurality of feature images recorded in the database 2 are supported, Alternatively, a setting is made so that a plurality of suspicious search targets appear simultaneously.

  Since the second video acquisition device 4 needs to move, and it is necessary to catch the direction change and the enlarged image, a movable telephoto imaging device is adopted. However, many of the general movable telephoto apparatuses have a universal base disposed on the telephoto apparatus and use a precise driving mechanism. Such a type is expensive and large in volume and heavy. Therefore, after use for a predetermined time, it is necessary to carry out maintenance at a high price due to wear. Therefore, the present invention has proposed a lightweight, active and low cost design. That is, the movable telephoto camera (see FIG. 2) employed by the second image acquisition device 4 is formed by disposing the movable reflector 6 at the front end of the fixed telephoto camera 5. The movable reflection device 6 (see FIG. 3) includes a base 7, a reflection mirror 8, a vertical drive device 9, and a horizontal drive device 10, and the reflection mirror 8 includes The rotationally movable fulcrum 11 on the back surface is coupled to the base 7, and the vertical drive device 9 and the lateral drive device 10 form a 90 ° sandwich angle via the fulcrum 11 (the sandwich angle is 180 ° or less). In the case of 0 degree or more, all can be implemented, and it is not limited to preferable 90 degrees), and is distributed and fixed to the base 7 and the tip part thereof is connected to the corresponding part of the back part of the reflecting mirror 8. With such a design, it is possible to provide a function capable of quickly converting the image catch angle. When such a design is adopted, the image catch angle can be converted quickly and precisely, and the precision and efficiency of converting the angle can be easily adjusted (distance between the reflecting mirror 8 and the fixed telephoto camera). Can be achieved by adjusting the control speed), and it is possible to overcome the problem that the machine speed does not reach the computer speed. In addition, the movable reflecting device 6 and the fixed telephoto photographing device 5 are separated from each other, and the movable reflecting device 6 has a simple, inexpensive, lightweight and only small volume, so that the movable telephoto device has a small volume. It can be said that this is a major breakthrough because the overall cost of the photographing apparatus can be reduced and the cost of extra management and maintenance can be reduced.

  The processing device 1 connects the database 2, the first video acquisition device 3, and the video acquisition device 4, which has a processor in which a computer, a server, a processing program is built in, and other information processing functions. Electronic devices can be used. The processing device 1 can receive images from the first video acquisition device 3 and the second video acquisition device 4, can control the catch operation of the second video acquisition device 4, and the caught image and the database 2 Compare and contrast with the corresponding feature image. In the present embodiment, the processing device 1 first performs an image identification operation on the wide-angle image that the first video acquisition device 3 continuously catches, and the first feature image (for example, in the wide-angle image) When a matching part appears in the stolen vehicle style shown in FIG. 4, the part is set to be an enlargement target, and then the second image acquisition device 4 is driven to enlarge the enlargement image of the enlargement target. To catch. Then, it is confirmed whether or not the enlarged image has a portion that matches the second feature image (for example, the outline of the vehicle tag shown in FIG. 5). And the previous one or other second video acquisition device 4 is driven to catch the enlarged image. Finally, if the difference between this enlarged image and the third feature image is compared and contrasted and it is found that it matches the third image feature (for example, the number of the tag shown in FIG. 6), the search target Assuming that a (stolen vehicle) has been found, at the same time as finding a search target, search reporting means (not shown) is activated to notify the police and promptly go to catch.

  The searching reporting means can be appropriately changed according to the scene where the automatic searching system of the present invention is applied. For example, when attached to a police vehicle, an alarm is issued and the number of the tag of the vehicle (last enlarged image) is displayed on the display device, or all matching images are displayed on the split screen, and the search is performed. By displaying the direction of the search target (stolen vehicle), the police in the vehicle can quickly identify the true search target (stolen vehicle) on the roadway and stop to perform reconnaissance. For example, when it is attached to a fixed location on the road, the relevant message is transmitted to the monitoring host device of the local police network as the searching report means, or the related information is transmitted to the local police network. The information is transmitted to the corresponding system for all police vehicles to be connected, and information on the related vehicle, the point information, the traveling direction information, and the like are notified to the police by display, voice, and other methods. The embodiment of the searching reporting means described here will be described simply, and the actual method can be suitably adopted depending on the application of the system or the difference in the search target, and they can be used later in the system. Needless to say, it is not limited to the above-described functions.

  In addition, the automatic searching system of the present invention can further include a display device (for example, a monitor) and is connected to the processing device 1 so that the wide-angle image and the enlarged image are displayed. Displays the actual image to be matched. Of course, it goes without saying that it is possible to track a continuous wide-angle image continuously by adopting two display devices, each displaying a wide-angle image and an enlarged image. It is.

  In addition, by introducing an operation interface (for example, control keyboard), the search target may be set to operate or the priority order of all search targets may be set, and the search target is found. Provide functions such as stopping the screen in some cases. However, the available functions of the operation interface are determined by the application program to be built-in in the processing device 1 and can be expanded with appropriate functions, for example, scouting the enlarged image of the suspected object manually, It goes without saying that the file of the database 2 can be loaded in, updated, etc., and is not limited to the above-mentioned use.

  In addition, the image captured by the system may generate a dynamic change due to the difference in relative speed, and this change should be overcome by the continuously captured wide-angle image of the first video acquisition device 3. It is. The second video acquisition device 3 continuously captures wide-angle images, and the part that matches the first feature image repeatedly appears at different positions in the continuous wide-angle image, so that the image is caught The relative movement amount and the speed can be produced according to the time difference to be performed and the position difference of the matching portion. In this case, the first amount of the angle to be changed by the second video acquisition device 4 can be acquired. This kind of dynamic image catch calculation has already been applied extensively to, for example, the above-mentioned patent image processing and attached technologies, and the calculation formulas and techniques are also points that appeal to the earliest invention of the present invention. Since it is not, a terrible explanation is omitted.

  In the case of explaining the automatic searching method of the present invention by taking as an example searching for a vehicle stolen by helping the police mentioned above, the method of the present invention has the following steps.

  Step a: Providing the database 2, the first feature image (for example, vehicle style) to be searched, the second feature image (for example, the wheel of the vehicle tag), and the third feature image (for example, the number of the vehicle tag). Record.

  Step b: Continuously catch wide angle image signals. The first video acquisition device 3 (for example, a fixed wide angle imaging device) continuously shoots wide angle images in a fixed video acquisition direction (see FIG. 4), and each wide angle image is captured. Sequentially transmitted to the processing device 1.

  Step c: Identify wide angle image. The processing device 1 is used to compare and contrast each wide-angle image 12 with the first feature image in the database 2 and match the wide-angle image to the first feature image (for example, in FIG. 4). If the vehicle has a contour 12), the contour 12 of the vehicle is set to be an enlargement target.

  Step d: Catch the enlarged image. The second image acquisition device 4 is driven by the processing device 1 to catch an enlarged image of the enlargement target (vehicle style contour 12).

  Step e: Identify the magnified image. The enlarged image and the second feature image recorded in the database 2 are compared and contrasted by the processing device 1, and the enlarged image (vehicle style contour 12) matches the second feature image (for example, shown in FIG. 5). When it is found that the vehicle tag outline 13) is provided, the portion of the vehicle tag outline 13 is set to be an enlargement target.

  Step f: Catch the enlarged image. The processing device 1 drives the front one or other second video acquisition device 4 to catch an enlarged image of the enlargement target (vehicle tag outline 13).

  Step g: Identify the magnified image. The processing device 1 compares and contrasts the enlarged image with the third feature image recorded in the database 2 and matches the third feature image in the enlarged image (vehicle tag outline 13). For example, the vehicle tag shown in FIG. Is found as having been searched for.

  Step h: Searching reporting means: When the searching object is found, the searching reporting means (the detailed description is omitted because it has already been described above) is started synchronously, and the enlarged image and the display device Display a wide angle image.

  In the above steps, the method for identifying the similarity between the image and the comparison can be achieved by a technique such as image identification software loaded into the processing device 1, and the digitized image identification technique is conventional. The description is omitted here.

  Of course, the automatic searching technology (including the system and method) of the present invention is not limited to finding a stolen vehicle in the police network described in the above embodiment, but various features. Any target that can be classified by the target can be a search target. For example, a specific human (for example, a missing human or a suspect) is selected as a search target according to features such as the face contour and the distance between both eyes. It may be defined and there are many examples that can be given, so a detailed explanation is omitted here.

  In addition, the technology for realizing the present invention is not limited to the two image acquisition devices described above, and a wide-angle image signal and an enlarged image are simply used by using the same image acquisition device (movable focus change telephoto device). The signal may be caught and a slight interruption phenomenon may occur due to the time difference, but it is still applicable to a case where the variation is small or slow (for example, parking lot). These variations are merely differences in the number of video acquisition devices, and all related searching techniques are small in difference from those described above, and will not be repeated here.

  As can be seen from the above description, the present invention can automatically catch a magnified image for any search target by combining the processing device and database design and method with the first and second video acquisition devices. , Which can provide clear images, perform step-by-step image comparison and discrimination, and since all the images provided for each discrimination are clear, there is no digital magnification blur phenomenon, Accuracy can be secured. Further, since a stepwise comparison method similar to human sensation can be adopted, searching efficiency can be improved. Compared to the prior art, the present invention can significantly improve the accuracy and efficiency of identification, and does not require any sensor or high grade image identification device, greatly reducing the cost of equipment It can be introduced to the use of each related industry, so it can be said that it is a serious breakthrough design. In addition, since the automatic search technology of the present invention is unprecedented, the present invention has novelty and inventive step of patent requirements and high industrial utility value. I would like to ask you to give me.

1 is a block diagram showing an automatic searching system of the present invention. It is explanatory drawing which shows operation | movement of the movable telephoto imaging device of this invention. It is a perspective view which shows the structure of FIG. It is explanatory drawing which shows the case where a wide angle image is caught by the 1st image | video acquisition apparatus of this invention. It is 1st explanatory drawing which shows the case where an enlarged image is caught by the 2nd image | video acquisition apparatus of this invention. It is 2nd explanatory drawing which shows the case where an enlarged image is caught by the 2nd image | video acquisition apparatus of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Processing apparatus 2 Database 3 1st image | video acquisition apparatus 4 2nd image | video acquisition apparatus 5 Fixed telephoto imaging apparatus 6 Movable reflection apparatus 7 Base 8 Reflection mirror 9 Longitudinal drive apparatus 10 Lateral drive apparatus 11 Support point 12 Vehicle style outline 13 Vehicle tag outline 14 Vehicle tag number

Claims (4)

A database for recording at least first and second feature images to be searched;
A first video acquisition device that continuously captures wide-angle images within the covered range;
A second video acquisition device that dynamically catches an enlarged image;
Connected to the database and the first and second video acquisition devices, performs image identification, and sets a portion that matches the first feature image of the wide-angle image to be an enlargement target. When the acquisition device is driven to catch the enlarged image of the enlargement target, and when a portion matching the second feature image in the enlarged image is found, it is considered that the search target is being searched, and at the same time the searching report means And at least a processing device that activates
Automatic searching system. A database for recording a plurality of feature images to be searched;
A first video acquisition device that continuously captures wide-angle images within the covered range;
A second video acquisition device that dynamically catches an enlarged image;
Connected to the database and the first and second video acquisition devices to perform image identification and first drive the second video acquisition device to enlarge the portion matching the first feature image in the wide angle image If you catch the video, catch the enlarged image of the part that follows and sequentially matches the other feature image, continue until you find the part of the one enlarged image that matches the last feature image, A processing device that considers that the search object has already been searched and simultaneously activates the searching report means,
Automatic searching system. Providing a database recording at least first and second feature images to be searched;
Continuously catching wide-angle images within the range covered by the first video acquisition device; and b.
Setting a portion matching the first feature image of the wide-angle image to be an enlargement target by performing image identification by the processing device; and c.
A step c of catching an enlarged image of the enlargement target by a second video acquisition device;
Considering the search target to be found if the processor finds that it has a portion matching the second feature image in the magnified image; and
At least a step f of activating the searching reporting means,
Automatic searching method. Providing a database for recording a plurality of feature images to be searched; and
Continuously catching wide-angle images within the range covered by the first video acquisition device; and b.
Setting the portion matching the first feature image in the wide angle image to be an enlargement target by performing image identification by the processing device; and c.
Step d for catching an enlarged image of the enlargement target by the second video acquisition device;
Sequentially identify magnified images that match other feature images, perform catch, and consider the search feature found when the last feature image matches the previous magnified image e,
A step f for activating the searching reporting means,
Automatic searching method.

Images