TWI220208B - Active search assisting system and search assisting method - Google Patents

Abstract

The active search assisting system and search assisting method primarily employs the first imaging device to continuously acquire the wide angle image signals within covered range and then let the processing device in connection with the database, the first imaging device, the second imaging device proceed image recognition to set up the wide angle image matching with the first characteristic image portion in the database to be enlargement target. Further drive the second imaging device to acquire the enlarged image of the enlargement target. Subsequently when recognizing the portion in the enlarged image matching with the second or final characteristic image in the database, the searching target is considered to be found and in the meantime activate a found alert mechanism.

Description

1220208 V. Description of the invention (1) < Technical field to which the invention belongs > The present invention relates to a co-seeking system and a co-seeking method, in particular a method using image recognition technology, and can perform a plurality of feature images of the co-seeking object. Advanced comparison and active search systems and search methods that actually capture enlarged images to improve clarity and recognition. < First method, and each characteristic person arrives at the license plate notification point or rate, but it is difficult to be too big. The technology is traditionally used to search for specific objects such as wanted criminals, stolen cars ... The human eyes are used to compare various features in the data, and most of these features can be presented as various images. For example, when searching for stolen cars, the data may include the model, car color, and license plate number. The search is often based on car color and model style as a priority comparison feature. When searching for vehicles with priority features, the license plate is further advanced compared. The number, if the number matches, it is deemed to be a stolen car for follow-up operations such as the owner's collection or dragging back to storage. Under the current situation, it is very ineffective for people to set jobs, which is not effective for fatigue, inattention, laziness, or the need to rest. The efficiency of finding jobs is not limited to the aforementioned search and comparison in the rapid development of computer and information industry motor vehicles High cost method, and personnel may affect the accuracy of identification, and even personnel can work full time. In general, in order to solve the high cost and low efficiency of the aforementioned comparison of users, in recent years, it has exhibited mature image recognition technology and applied it to some specific recognition applications. For example:

Page 5 1220208 V. Description of the invention (2) I. A kind of diagram proposed by the company in China's Announcement No. 3 9 9 to identify the type of figure. 2. China's Announcement No. 406 Co., Ltd. Telecommunications mode-based vehicle. The vehicle is tested from the front to the top with the image recognition technology. 3. China Announcement No. 422 Co., Ltd. proposes to use cameras to capture vehicles for image comparison and identification. 4. China Announcement No. 512 Co., Ltd. proposes a system that mainly uses (vehicles must be stopped in the software) to identify license plates and other public facilities. Vehicles in the transportation facilities of the vehicle type 18 Fractal identification 2 5 After studying the vehicle type, 9 6 a type of image recognition, 2 8 multiple photographic ground codes, and the invention patent for the tube license plate No. 8 are used by the United States Hewlett-Packard Co. According to the invention patent No. 0, it is a hidden Markov recognition system proposed by Chunghwa Telecom. It mainly uses the range finder's height model to convert to the vehicle's appearance, and recognizes the vehicle's No. 1 license plate based on the vehicle's height value. After the identification, it is equipped with the 8-point video camera shooting point), the supply system, the identity management system, the patent recognition system, and the division of the vehicle. The license plate is used from the vehicle for reuse in the road work fee system. The method is based on the method and the binary license plate number, which is identified by China Dynamics, the previous license plate image, the road, the bridge, the vehicle management system, and the main code. Telecommunications vehicles and post-technical technology, technology or technology, and technology control technology license plate control technology (parking wisdom from the previous disclosure of the patents before the case and technology can be summarized, image recognition technology has become mature and can be widely used, but in The current practical applications are identification of fixed patterns at close range, or the need to manually enter comparison data in real time.

Page 6 1220208 V. Description of the invention (3) Movement mode. Because it is even necessary to be a signal, and further, it is difficult to recognize after zooming in to achieve the difficulty of replacing the suspect blur phenomenon in the field. Furthermore, the technique can be passed through the image. Yet technology. The time provided by the right or the unit will be lost, and the good will still need to be added. The patents will be identified, and they will all be displayed, even the active image. Although it is identified, it is not a major assistance to go to the case. There are many previous cases, but it is necessary to recognize that the face details are ambiguous than all the cases. Strengthening the proactive technology to replace different external benefits is no matter the number. With the help of close-up features, some are clear to the country of or? The function of the human and the use of the capture position to place artificially, magnifying features of other homes in the country, or the image-like sense of damage caused by the owner of the large-scale image of the model, through the, or single-unit approval, But the public can, the maintenance of a single image of the mobile searcher is more difficult than the inspection of the monitor often to identify the position to solve the problem. Both parties can only find special auxiliary negative image re-images, yes, the viewer Because there is film reading and preparation of these documents, there is a fixed purpose to provide the burden. Proceed to the south, so Lu Jilu digitally zoomed in on the image strength of the industry, a few round trips, watched the flowers, restricted the range of the target work, and various actions. The change in the degree of help required by the country that the image needs to be enhanced by the Chemistry Division < Inventive Content > In view of the fact that the above-mentioned conventional techniques using image recognition cannot provide the effect of actively searching for a specific target, and the captured image is enlarged It is also difficult to effectively identify such shortcomings. The inventor then tried to use the existing feasible image recognition technology to cooperate with the active acquisition of magnified images of a plurality of characteristic images matching the object of the search

Page 7 1220208 V. Description of the invention (4) The image provides advanced comparison to overcome the shortcomings of the conventional technology, and provides an active co-seeker who can actively capture clear images for identification. Therefore, the main object of the present invention is to provide an active search that can perform advanced comparison on a plurality of feature images of a co-seeking object, and actually capture enlarged images to improve the clarity and recognition effect. Systemist 〇 Another object of the present invention is to provide an active method that can perform advanced comparison of a plurality of characteristic images of a co-seeking object, and actually capture enlarged images to improve clarity and recognition effect. Cooperative methods. In order to achieve the above-mentioned system, at least images are included. The main purpose of the present invention is to realize an active cooperative search database, which records at least first and second features of a cooperative search object, and a first image capturing device for continuous capturing. The wide-angle image within the coverage range; the image pickup and processing device can be used to identify large eyes, and the time is set to set the shadow part as the target's magnified shadow feature image alarm mechanism. Position to dynamically capture the magnified image; and to connect to the aforementioned database, the first and second image capturing equipment and match the wide-angle image to the target of the first feature image, and then drive the second image capturing device to capture the magnification Continuing in identifying the magnified image, it is consistent with the second being regarded as finding the co-seeking object, and simultaneously starting a search to achieve another purpose, the present invention is implemented in an active co-seeking method,

Page 8 1220208 V. Description of the invention (5) includes the following steps: a. Provide a database to record at least first and second characteristic images of the co-seeking object; b. Use the first image capturing device to continuously capture Wide-angle images in the coverage area; c uses a processing device to perform image recognition to set the wide-angle image to the first feature image as the magnification target; d uses the second image capture device to capture the magnified image of the magnification target; e. using the processing device to identify a part of the enlarged image that matches the second feature image, it is regarded as finding the co-seeking object; and f. starting a finding alarm mechanism. When there are more than one difference in the number of cooperative search pairs, in order to reach the cooperative search system, at least one database, a first image acquisition; a second image acquisition, and a processing device are arranged to match the first image in the first image. Identify and capture the characteristics of its appearance. Therefore, the above mainly includes: for recording the device, using the device, using, connecting to the identification and the first characteristic shadow image, depending on the actual object, when the characteristic image of the co-seeking object exceeds two purposes, the present invention A plurality of characteristic images realized in an active cooperative search object; continuously capturing wide-angle shadows within a coverage range to dynamically capture enlarged images; and the aforementioned database, the first and second image capturing devices driving the second image capturing first The device captures an enlarged image of a part of the wide-angle image, and subsequently sequentially enlarges the image as an anastomosis part until it is recognized.

Page 9 1220208 V. Description of the invention (6) When the last feature image matches the part of the previous enlarged image, it is regarded as finding the co-seeking object, and a finding alarm mechanism is activated at the same time. Similarly, when there are more than two feature images of the search object, in order to achieve the above-mentioned another object, the present invention is implemented in an active search method, which includes the following steps: a. Provide a database for recording the search Multiple characteristic images of the object b. Using the first image capturing device to continuously capture wide-angle images within the coverage range; using a processing device for image recognition to set the wide-angle image matching the first characteristic image as the magnification target ; Use the second image capturing device to capture the magnified image of the magnified target; sequentially identify and capture magnified images that coincide with other feature images until the last feature image that matches the part of the previous magnified image is identified It is deemed that the co-seeking object has been found at any time; and ί. Start a discovery alert mechanism. < Embodiment > In order to make your reviewing committee aware of the purpose, features, and effects of the present invention, the present invention will be described in detail through the following specific examples and the accompanying drawings, which will be described later. Please refer to the first figure, which is a schematic block diagram of a system of the active cooperative seeking system of the present invention. As shown in the figure ... The active cooperative seeking system of the present invention includes at least a processing device 1, a database 2, a first image capturing device 3, and a second image capturing device

Page 10 1220208 V. Description of the invention (7) Composed of 4; The database of the electronic device is hard to find the characteristic image of the object. Let's take as an example a model or color phase outline. In the third example, the feature image of the present invention and the second. It is necessary to pay attention to the levy image, not the co-seeking object. Add more co-finances. The first image is taken. The four camera capture angles are used for the second set and used for the machine. Although the image is set in the camera, Several optional dynamics of image acquisition should be set as the capture device, the second one of which can be disc, the first one, because of the appearance, the characteristic data is specially, the record is selected as the capture. Two special cases for enlarging the built-in memory, and the image library 2 applies for the three records, and the special 3 is a solid-magnification image for the target learner. The rtrt image is installed in, for example, the body or its image. The first and second images can be used in the car and the second database. The data image is used for fixed wide image. The magnification of the 4 series is so juxtaposed by the number of images taken here. It is used for recording electronic recordings, recording the three special Zhongyi image libraries, 2 brains, and 2 recording media for video image code. The feature can be selected to be in the middle or more body, and the feature shadow can be used to assist the police image. Among them, the object-finding image is an optional record, which is also not available, or can be used for image and third-party search. If the thief is the first to record the license plates of three stolen cars, this is a special example. There are more special records to record an update at any time or continuous angle capture photography. image. No longer set to 3 and yes, this should not be used to correspondingly get the culprit. It is chosen to be equipped with one of which is redundant, and the first is limited to a wide angle. The second image capturing device is a movable telescope set to 1 for use because of the fixed type. In addition, the second image capturing device 4 and the image capturing device 3 of this embodiment may be provided with a record in the aforementioned information.

Page 11 1220208 V. Description of the invention (8) The characteristic images in the library 2 or make up for the simultaneous appearance of multiple suspected co-seeking objects. Since the second image capturing device 4 must be moved, a movable telephoto camera is required. However, after telephoto shooting, this form is often used for sharpening and mentioning the second reflection. Figure 3 and one side can be driven horizontally to a maximum of 100.) Mirror 8 back and horizontal to change speed and efficiency. You can reach the calculation speed of a second picture) Device) Then the driving of the 80-distribution surface of the horizontal rotation actuator is solidified and the precision (into) of the second-moving camera is not only damaged but a kind of light image, which is included in 6 The use of 10-degree driving fulcrum and fixed-type counterpart 1 is fixed. The price of the camera-mounted style is expensive, and the device consists of one device and one device. The 11 11 system is larger than the base. 0 and the telephoto image adjustment can be repeated again and again, said ΦBei Fei amount as low as 4 fixed 〇 the base 0 combined with the fulcrum zero degree seat 7 to make the anti-photographic angle reflection set by the camera must be universal, and The external version adopts the movable type 7, which is composed on the base 11 and is designed with a mirror lens machine 5 and a conversion mirror 8 meter. It can be mounted on a base and has a large volume for maintenance and design, such as a movable telephoto reflection-reflection. Ninety can be implemented and made on the seat 7, which can pass through the image produced by the 8 ', and is easy to fix and easy to move in the reverse direction. It can be coordinated and repaired under the following conditions. For the degree clip, the front end is not controlled by the counterpart. The precise drive mechanism of the action telephoto camera, which is located in the modulation mode to overcome shooting equipment and capture magnified images, takes a period of time to use. Therefore, the telephoto front end 6 (, and the longitudinal angle of the mirror (are deflected by the vertical direction respectively. According to the setting of the telephoto mechanical camera 6 camera settings, please install a vertical 8 series drive with a good connection angle to the drive, otherwise set the angle shadow. The speed of the machine is the same as that of the motor (see the movable combination of the driver for its backing device 9 and the angle between the ninety-degree and the reflector 9 shooting angle meter.)

Page 12 1220208 V. Description of the invention (9) The telephoto camera is cheap, and the overall cost is broken. The images captured by the processing unit 3 and the processing unit device 4 of the first processing program. In the three consecutive images of this implementation, the car that kisses the stolen car appears and drives the second one to further distinguish (for example, the fourth feature image of the imaging device in the second part, such as the sixth stolen car). The alarm mechanism (the aforementioned 5 has a small mutual product, and the device 1 can be accessed by the second image acquisition processor 1, and the control image is the same as the type captured in this embodiment). It is set to capture the similarities and differences as shown in the figure. When you do not plan to find the light, separate the device, or collect the system data, a large part of the wide angle is set to 4 large kisses to enlarge and enlarge. Obtained), the structure of the police plane and the movable reflection device 6 is simple, so the maintenance and repair costs outside the movable telephoto camera can be relatively reduced. It can be said that it is a sudden jump to 4, and when the image features are processed from the second library 1, The outbound number of the unified target image in the captured image should be made in order to make the system. The database can be used to image the device in the first image pack and put the license plate on it. The most recent kiss), the pairing notice should be regarded as a material acquisition, 4 should be 1 series of images (for example, some large heads have contour driving, and then identify the video when the main net 2, the computer, the processing, The capture image of the functional device 3 is first matched to the identification unit (such as the fourth setting as the target). The third recognition is the same. The shadow of this action is shown in the first work, the second copy of the big shadow enlarged, or another big shadow feature. This step forwards to the pursuit of the image pickup device, and the second sub-image is used to identify the image of the target. If there is an image and image matching action. System built-in. Take images and enter into the film and film association, and then enter the film to match the second and third time (image (find application)

Page 13 1220208 V. Description of Invention (ίο) Occasion varies. Example 1: The display device display is divided into other information such as security clearance, local police, etc. on the road because it is and belongs to (for example, display display device ri-Jl. R ± > 5) It also has function expansion and more time The way to find the right one is to install it in the Internet. In reality, it is also based on the wide-angle monitoring of the Daying device, and then the association looks at the end, such as the appearance of new assets. (There is a local police notification on the road of the association. The system does not use this screen image and other individual images. It also seeks to recognize the processing, such as hand storage, such as displaying the stolen car, and finding the police officer on the fixed police network. Wide-angle mounted on the license plate display) The image of the location (the location of the prison is related to the description of the search engine is connected to the master of the image of the police car number to identify the location of stolen vehicles, the control of the host vehicle received the report after the paragraph is When the real-time motion-restricted processing is put on (the most savvy, to), the search function corresponding to the information police application function can be combined to make the car and then get the system or the system, the local system only or the cooperation (commonly as Send a magnified image, the police officer will conduct the police blocking mechanism including the transmission, the point information is a feasible target, often people are exhausted, the image can be stopped quickly, the relevant display can be related to the actual operation of the aircraft, and), Includes speed detection and detection to send information, and the driver to act in the same way, or to display the car through the display. If you send the photo to the sound or the information, the difference, the reason) It can include an operation icon or set all the targets The day-to-day device 1 includes an active detection system for detecting suspicious pair 2 cases, and the like, and the search system may further include a display device device 1 for displaying the wide-angle image. Of course, it is also possible to use two large images to facilitate continuous tracking of continuous interfaces (such as the control keyboard), to operate the priority order of the coordinated objects, and to freeze the functions. However, depending on the software used in the operation interface, it can be used to enlarge the image of the appropriate function, or used to load, and is not limited to the aforementioned purposes.

Page 14 1220208 V. Description of the invention (π) In addition, different continuous shooting can be used for continuous shooting. The time difference between the current time and the current time can be different. Because of this kind of image processing and creation or appeal, the previous method of discovery is great. A. Provide the image outline. B. Continuously send the solid-angle image to c. Identify the image. It can be relatively transformed to Zhu Qiao. Because of this dynamic variable wide-angle shadow angle image, the wide-angle coincidence is used to calculate the dynamic follow-up technique point. It is used to synchronize and include a piece of data (for example, and to capture a wide-definition wide image (such as processing wide-angle Shadow 2 and this, Dang Guang (as converted by the Twelve 12 system, this image is added, and the part of the image is in the second image acquisition technique, this will no longer help the police the following step library 2, models) The fourth image of the special angle image angle photography is set to 1; the image is the change of the part of the four images of the angle image of the database. The image device has been widely calculated in order to overcome the difference in the current position. Step: Used for recording, the second sign of the video signal, and the machine) (shown even)) The car with the display in the record at the use point 1 must be used for the video. Due to the first preference, the calculation of 4 should be general application or technical description. The car is taken as an example, because the part of the phase-converted image is taken by the angle of the movement amount, as the previous one is not set to the speed of 3 The 3 series will re-take the image and the speed of the amount of money, the first feature image of the active recording of the object of the present invention (such as the license plate wheel (such as the license plate number); use the first Imaging device 3 (continuously captures a wide range of fixed image orientations, and sequentially processes each wide-angle image. Device 1 matches the contour of the first feature image of each wide-angle video to the contour of the first feature image. 1 2 ), The big goal of the car;

Page 15 1220208 V. Description of the invention (12) d. Capture 4 captures e. Identify and deserve this special feature) f. Pick up the first special feature) g · Identify and deserve this special feature) h. Start and start borrowing By like. The foregoing steps are permissible, and the numbers are explained separately. Of course, zoom in and take the enlarged data to enlarge the image, and then zoom in again to get the enlarged data to enlarge the image, that is to find a picture such as the image, the magnified image, the image in the library 1 (the part of the image is the image of the car, the image of the device) ; Image, the image in library 1 (the part of the image is regarded as the alarm-seeking machine and the alarm display is not installed by the target (by Lu Jilu model copies) The processing takes the outline of the first license plate recorded by the department (if the cooperative search system is obtained, when the search mechanism (such as the display step is used, it is used to compare the imaged image contained in the management device 1) The recognition technology device 1 drives the enlarged image of the second image capturing device contour 12); the processing device 1 recognizes the enlarged image and compares the two characteristic images, and 12) has the license plate contour 1 3 shown in FIG. 3 parts are set as the magnification target; the device 1 drives the same or another magnification target (the license plate outline 1 3 the physical device 1 recognizes the magnified image and compares the three characteristic images, and 12) has a match with the third The license plate number 1 4 object shown in the picture. When the object is found, you can synchronize the above, no need to repeat it), and the magnified image or the method including wide-angle shadow, identification image similarities and differences, such as identification software to achieve the system Those skilled in the art are not aware of the active cooperative technology (including systems and methods) of the present invention.

Page 16 1220208 V. Description of the invention (13) is not limited to the use of the police network in the previous embodiment for pursuing stolen cars. Any subject that can be distinguished by characteristics can become a search for the applicable scope of the present invention. Face contours and eyes are defined to search for a specific person (such as missing persons or wanted persons). The examples are so numerous that they are too numerous to list here. Furthermore, the implementation technology of the present invention is not previously disclosed. The device is limited. You can also use the same image capturing device (movable camera) to capture wide-angle image signals and enlarged image signals. Although there may be slight interruptions, it can still meet some small variations or slow trains.) Occasion. Because these changes only differ in the search technology related to the number of image capturing devices, they are very different from the previous one, so it will not be repeated from the foregoing description. The present invention uses the first and second processing equipment and database design and method. The application can actively seek out the enlarged image by looking for the object, so as to improve the details and provide advanced image comparison and identification. Since each provided identification is clear and legible, there is no risk of digital magnification and blurring. Identification accuracy. In addition, because of the anthropomorphic approach, search efficiency can be improved. Compared with the conventional technology, it can obviously improve the accuracy and efficiency of identification, and it does not need to increase the number of detectors or use high-end image recognition instruments, so it is greatly reduced, which can be beneficial to the application of various related industries, which is a major one. Moreover, the active search technology of the present invention has never been seen before. It should meet the novelty, advanced elements, and high-level industries of the invention patent. For example, all kinds of images constitute characteristics such as distance, etc., and the focus time is two parts (for example, the amount of details will be described in detail as a result of the order of Ren Qing's understanding of Ren Qing). Raise, 〇 set up the Hexie video image can really compare the sense of invention cost design invention value

Page 17 1220208 V. Description of the invention (14) Although the present invention has been disclosed above with a preferred embodiment, it is not intended to limit the scope of implementation of the present invention. Anyone skilled in the art can make some changes and modifications without departing from the spirit and scope of the present invention. That is, all equal changes and modifications made in accordance with the present invention shall be covered by the scope of the patent of the present invention. The definition shall be based on the scope of patent application.

Page 18 ......... 0 12 3 1234567891111 1220208 Brief description of the drawings The first diagram is a system block diagram of the active cooperative seek system of the present invention. The second figure is a schematic diagram of the action of the telephoto camera of the present invention. The third figure is a schematic diagram of the three-dimensional structure of the second figure. The fourth figure is a schematic diagram of capturing a wide-angle image by the first image capturing device of the present invention. The fifth diagram is a first schematic diagram of an enlarged image captured by a second image capturing device of the present invention. The sixth diagram is a second schematic diagram of the present invention using the second image capturing device to capture an enlarged image. Brief description of drawing number: ..... processing device ... Database • • • First image pickup device • • • Second image pickup device • Fixed telephoto camera •• Moving reflection device ... .... base ... mirror • ·· • longitudinal drive • ·· • transverse drive ............ pivot point ..... model profile ..... license plate profile

Page 19 1220208 Schematic illustration License plate number 14 · 1111 Page 20

Claims (1)

1220208 6. Scope of patent application 1 A main data image; a dynamic search system library for recording, at least including: at least the first and second characteristics of the search target, a first image capturing device for continuously capturing the coverage The wide-angle image inside; the second processing image pickup device, which sets the large target setting for the image, takes the part of the image and matches it with the second and simultaneously starts 2 kinds of main data and a first image; a second one processing Positioning, sequential images with angular images, straight part of the mechanism. The dynamic cooperative library uses an image acquisition device to record the image acquisition system with a magnified feature image search system that is connected to the image recognition as an enlargement. The image acquisition device and the device are used to connect the image recognition to the first identification and When capturing and identifying the last one, it is considered to find the database that dynamically captures the enlarged image, the first, and the wide-angle image is matched to the target, and then the second image is driven, and the identification of the playback part is continued. Seen as a police mechanism. , Including at least: a plurality of co-finding objects to continuously capture the coverage range to dynamically capture the enlarged image, the aforementioned database, the first and first driving part of the second image feature image, and other feature image anastomosis features The image coincides with the previous co-finder object and is started at the same time; and the second feature captures the first feature image capturing device to capture the symbolic image of the co-finding pair in the large image; the wide-angle shadow inside; and the second image capturing device capture Take a large number of images, and the next enlarged image to find an alarm Page 21 1220208 VI. Scope of patent application 3. An active co-seeking method includes the following steps: a. Provide a database to record at least first and second feature images of the co-seeking object; b. An imaging device continuously captures wide-angle images within the coverage area; c. Uses a processing device to perform image recognition to set the wide-angle image matching the first feature image as a magnification target; d. Uses a second imaging device To capture a magnified image of the magnified target; e. Use the processing device to identify a portion of the magnified image that matches the second characteristic image, consider it to be the target of the co-finding object; and start a search alarm mechanism . 4 · An active co-finding method, comprising the following steps: a. Providing a database for recording a plurality of characteristic images of the co-finding object b. Using a first image capturing device to continuously capture wide-angle images within a coverage range; c. using a processing device for image recognition to set the wide-angle image matching the first feature image as the magnification target; d. using the second image capturing device to capture the magnified image of the magnification target; e. sequentially Identify and capture the enlarged image that matches the other feature images until the last feature image that matches the part of the previous enlarged image is considered as the co-seeking object; and f. Activate a search alarm mechanism . Page 22