TWI338504B - - Google Patents

Description

1338504 IX. Description of the Invention: [Technical Field] The present invention relates to an intelligent monitoring system; and more particularly to a video analysis and storage method required for monitoring through a surveillance camera. [Prior technology] Whether it is financial _ offer often there will be a big 4 cash touch, Daxiang will install a handle machine, - the side will frustrate the culprits not to have a non-participation, another - _ provide police ^ in the ^ to do the case data of. Just, the general camera _ fixed position, shape shooting angle to shoot, Shu Cai coffee grain set a specific fan _ to _ photo, so for those who have the heart dodge reduction, you can still replenish the complete (four) image. In order to overcome this problem, the manufacturer has proposed the Chinese Patent Publication No. _ No. _ "Storage and Turning Device Based on Image Age Objects", and by the pursuit of the situation _彡Monitoring, especially for moving objects such as people. In order to be able to monitor the location of the video for the video surveillance, in addition to the tracking technology itself, it is more necessary to have the "No. With this camera, after receiving the relative control command, the __ shadow can be made vertical and horizontal, and seamless, no 5 dead angle shooting action. = If you can add reading control, you will change it. To this end, the Republic of China Announcement No. 46: Objects go into two: for certain characteristics, and then allow or disallow specific super-business people must take off the cow earning that 'for the forced mother to enter, that is, Lai (four) (four) exists," When the employee does not take off the *full cap, the automatic Η will not open. In addition to the above application of tracking, identification, etc., the remote monitoring is also included, the remote monitoring is mainly taken by the camera. Video surveillance domain, and then monitor the host to view the material _ money to the camera's visual HI through the & line network to travel to the table end, mobile devices. Such as the Chinese people, the number (4) G53_4 patent "silk image monitoring system The main purpose is to store video images of fixed time before and after the intrusion of objects, and to provide users with the Internet to obtain the current image. In addition to the above network monitoring, it is also possible to transfer a higher GPRS to achieve remote monitoring purposes. For example, the Republic of China Announcement No. M287021, "Integrated Packet Radio Service (GPRS) Supervision and Shadow Device", which transmits video data of intruding objects recorded by the main camera, is transmitted to the user's mobile phone through the integrated packet radio service (4) S). And let the user use the SMS service (SMS) to send SMS control commands to control the surveillance camera. In this prior art, although SMS can be used to control the face machine, it is impossible to transmit the frequency through the SMS transmission command, and the amount of the image is invaded by the viewer, so that the system can not be monitored immediately. Unable to (4) monitor the title 'such as the Chinese coffee announcement number _ threat number ^ ^ turn over the blame (four) narration", which was transferred to the short-cut video... 彳 video read 'transmitted to the network value through the wireless sound, the user Through the Internet scales to turn over the post to get monitoring, _ provided by the "face control surveillance camera to achieve real-time monitoring.

_ 'Through __ road or 3G/GPRS for remote monitoring, there are many conveniences and benefits', but whether it is 'Na scale or 3G/GpRs, the width is still the same as the limit = monitor several field scales, no All of them happened to be off, and at the same time they were all passed at the same time: like the time, 'The Valley is easy to receive any images or messages because of the line. Play the above video images in sequence with A to avoid · and can't receive any images or news. But in case the video that really has an intrusion happens, it is ranked in the order of comparison. When the system is formed, the invaders may have left the scene.

In order to solve the problem of remote monitoring through the Internet or 3G/GpRS, that is, to minimize the use of bandwidth 4, (10) it is necessary to shoot (when the sugar is invaded), then use the bandwidth to achieve true Instant monitoring. In addition, in the present invention, the above-mentioned characteristics such as tracking, remote control camera, face _, etc. are still used to enhance the monitoring capability of the system of the present invention! > 1338504 SUMMARY OF THE INVENTION The main object of the present invention is to provide an intelligent monitoring and analysis system and method for pre-storing images or images of suspected intruders to provide relevant parties (for example, house L) to determine whether or not It is true that some invaders (such as thieves) will not cause unnecessary panic due to computer misjudgment. The invention has the following advantages: the purpose of providing a kind of intelligent monitoring and monitoring (four) female method, in order to monitor a large number of places, in order to reduce the burden on the bandwidth, the silk transmission phase: the image of each place (less In the case of a user who confirms that there is a suspected thief, the previously stored intruder image is played for further confirmation. (1) Based on the above purpose, in the intelligent monitoring and analysis system and method of the present invention, the wheel corridor of the suspected invading object is first judged and extracted, and then the temple is continuously tracked and continuously photographed through the camera, and obtained. The film is temporarily stored, and finally through the various communication channels 'to the clearest picture (occupying less bandwidth) in the above video, the master of the game, if it is determined that the suspect is a thief, only occupy the multi-bandwidth to actually download and play the film. The advantages and spirit of the present invention will be further understood from the following detailed description of the invention and the accompanying drawings. - Inch [Embodiment] In the intelligent monitoring sub-group (four), system and method of the present invention, it can be roughly divided into a clear transmission face (that is, a person who invades a person), a transfer place fine t_a film containing an intruder, The remote camera position is a large part of the r_fu 4 _, and the UJ8504 process will be described in detail below. > Read the first 1β map '1A, which is a schematic diagram of the transmission of the clear face of the present invention. ^ The intelligent intelligent monitoring system H of the present invention mainly includes a camera, a human intrusion judging module 16, a _ chasing module 18, a video database 2G, a clear painting group 22, a squaring group 24, and a county. When it is perceived that a person invading the invaders invades the object, T transmits a relatively clear picture to the client device 12. Among them, the camera 拍摄 shoots the predetermined place and obtains a series of multiple shot frames. When the right side of the day is riding a number of green monitoring places, the opposite camera 14 is installed.琢 2 ' 本 本 本 本 本 本 本 本 本 本 本 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏 魏The object wheel of the suspected person invading the object will be captured. When the tracking group 18 is connected to the contour of the object, the contour can be continued according to the contour and continuously photographed through the camera 14 and the lining is temporarily stored in the image data. In the library 20. Finally, through the various possible communication channels (such as the Internet, 3 main G/GPRS, fax (four)' will clear the face module 22 from the top _ out of the most = picture (occupy less bandwidth) For example, the Qing file) is transmitted to the client m (for example, when the object is suspected of invading the object, the image is first subtracted (for example, subtracted from the empty scene image), Obtaining suspicious objects after image binarization ~ like 'reuse morphology (m〇rphc) 1〇gy) to remove image noise and use 1338504 CCL (C〇nnected c_fiber t (four) one) and minimum rectangle _) The contour of the above objects. When the frame contains intrusive objects, the frame of the pens related to the invading objects will be started: the location will be recorded and the object will not be invaded until the place is difficult. By cropping, feature extraction, feature analysis, etc. on the frame of the site, it is difficult to find out the proportion of the frame occupied by the contour of the object and/or the wheel age of the object. The opposite part of the face of the face, and the relative position of the frame money into a clear picture. People such as human face, height, etc. 'must be able to easily pay from the clear face (or key face) enough Whether the owner is a thief is a clear picture. For example, the license plate image must appear in the image for clarity. In addition, the other method of clearing the face module is to let the parts of the object Can be clearly displayed. / Human example - such as the head, hands, lying, feet, etc., the back of the clothes pattern, etc., / member out _ key 昼 face; take the car as an example, the front and back images, covering the car Lamp shape, head shape The scales on the rear of the car, etc., are included in the face towel, which will help to establish the index dragon of the car by manual or automatic image analysis. Since the front side of the person and the front side of the cow cannot simultaneously appear in the image, The key is clear 昼°, not /, U. In principle, whether the object is clear or not is affected by three factors 'ie camera, environment, and object. The resolution of the camera, the adjustment of the aperture of the aperture, etc.' will affect The object is more sturdy. Wei's bright ambiguity ^ is enough, is 1338504, there is background #? Covering objects' will also affect the clarity of the object ^ In addition, the moving speed and posture of the object itself is the slave face to the photography, it will also clear In order to achieve the clear picture of the above-mentioned age, this technology rides the characteristics of each image of the segment---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Six categories of objects, such as front and back detection. Analysis of the degree of the object, can be measured by the wealth (Edge

Techniques such as Detect ion or High-pass Filter extract edge or high frequency information from the image of the object to analyze its blur. Object texture analysis can help extract patterns from objects, such as High-passFi 11er or Markov Random Field. Object color analysis allows you to capture color information, such as skin color (9) color, by Glustering or Gaussian Mlxture Models. The shape analysis of the object is performed by analyzing the shape information of the object, such as length, width, axis, wheelbase, area, and wheel scale, by a general statistical method or by Active Toitoui. The dynamic analysis of the object helps to understand the direction of movement of the object, which can be done by Bl〇ekMatGhing, optical flow method ((4)(7)丨F1〇w). The object is analyzed by face analysis, which can be done through the face _, the car off measurement, and the moving direction of the object obtained by dynamic analysis. According to the above six analysis methods, the characteristics of the object are obtained, and the key surface is selected by maximizing the expected value, as shown in the following formula:

Keyframes = arg max 0 (/,, Λ, /3, Λ, /5, /J Framet) 1338504 where /, Daisi Lai _ degree analysis - the amount of listening, moth table object color analysis / / 3 generations of silk analysis of the analysis of a group of special miscellaneous 'money wire pieces of the job - the amount of the transfer value of 4 represents the object of the analysis of the analysis of the group of quantitative feature values, the mechanical object is on the back of the detection - The group quantifies the special; the pound tear represents a certain number of shadows (four) / 1 knowledgeable age) represents the score from a certain Wei _, and calculates its score for the match. _ represents the few points with the highest scores as a key picture. The expectation function (X) can be calculated by the common linear weighting method:

叫,Λ,/3,以,Λ) = M W (10) The feature with a larger weight value in the secret is the more important it is to the sharpness of the image. The public ^ can be further extended, not only a set of quantized feature values have a weighting value, but also the parental feature values have a weighting value of (10): 〇C/i, -Wr_,) = W+... + v^ where /) ". Λ is the total of the measurable eigenvalues obtained by the six analysis methods. After the complete analysis = ΤΙ analysis and the expectation of the maximum value, the clearest surface can be obtained. Sometimes in order to carry out:: A more detailed object is clear and kneading, you can carry out "object cutting: two two:::" and analyze each part of the color, the color of the news, the square wire will seek the head of the head. The mouth ". _ sign, this feature can be used as a eigenvalue of the clear face function command 12 1338504. Take the car as an example of 'the car head, the car, the tail, the baggage equivalent parts' and analyze the various jobs Bribery, it is more clear than the degree of ambiguity, to determine the clarity of the image. As shown in Figure 1B_, if the invention has several places to be monitored, the above-mentioned camera 14 can be installed through each location. The clear pictures are all transmitted to the client device 12. The client device 12 transmits the display mode. 34 cut the facet displayed on the display 3〇 into the opposite block coffee ~ 30d, in order to simultaneously present a clear picture from different places. So - to 'except by the shadow record library 2Q pre-aged suspected invaders The image or image is provided to the relevant person (such as the homeowner) to determine whether there is an intrusive object (such as a thief). 'There will be no unnecessary confusion caused by the computer misjudgment. Even if it is necessary to monitor the location of the place, only the relative information is transmitted first. The image of each place (compared to the frequency of seeing) 'and then the user seems to have a small correction, the new inferior age in the image database 20 of the invaders image 'for further step-by-step confirmation. Not only can it reduce the burden on the bandwidth, but also can monitor several places at the same time. Before explaining how to play the intruder image stored in the image database 2G beforehand, please go back to the means of passing the county. Each location's camera 14 and the incoming (four) 嶋 group 丨 6 and / or the _ group 丨 8 can be transmitted over a long distance Internet or analog slow line (wired or wireless), while sending Module 24 and user There is a different conduit between the devices 12 relative to the client device 12. If the client device 12 is a mobile phone supporting GPRS or 3G (for example, 13 1338504 GSMC Global System for Mobile Communications) >CDMA (Code division multiple Access) ^WCDMA (Wideband Code Division Multiple Access) when the pipeline is GpRS or 3G network, so the relative module 24 can be used to use legs s (Multimedia

The Messaging Service) delivers the above-mentioned clarity. If the client device 12 is a computer device supporting the Internet and the browser, the pipe is the Internet, so the transmitting module 24 can transmit the above-mentioned clear face by using the TCP/IP protocol.

However, 'do not know which kind of official channel or means to send a clear picture, you can pass the intrusion message when you close (for example, the system finds the suspected intruder, please pay attention!), and attach the location video address relative to the clear face. ((4) (10) (10) Iron L (10) 攸, ancestor) and the place of interest depend on the body, and the continuation of the supplier from the display of the (. It should be noted that if the case itself is directly transmitted, the screen is converted into a format supported by the client device 丨2 by converting the group 26 as in the second figure. Please refer to (4) (4) '2A ~ (4) Schematic diagram of the book. As shown in FIG. 2A, the present invention is mainly composed of =1, the intrusion determination module 16, the _ ranching group 18, the image traits, and the monthly stencil module 22 The sending module 24 further includes a format conversion module 26. After the right user from the display 30a as shown in FIG. 1B, the present invention wins the 1〇gp'#-face, selects the block in the upper left corner. (4) will convert the mode video broadcast through the format (relative In the block business, the relative location of the site is converted into a variety of formats (eg 1338504) such as 3GPP (Third-Generation Partnership Program), 3GPP2 (Third-Generation Partnership Program 2) OMA (Open Mobile Alliance) ), to provide the transmitting module 24 to actually transmit the location video file to the opposite client device 12, as shown in FIG. 2B. If the format conversion module 26 starts the format conversion after the user selects the block, the problem often arises is that the system has to wait for a long time to watch the video.

Therefore, in another method, after the side tracking group 18 and the Qing Lai group 22 respectively perform the tracking video and the clearing surface, the grid capture frame is immediately converted into the video frame and stored. stand up. In this way, after the gamma selects the block 30a, the converted video clip of the object can be played immediately. If the client device 12 is a mobile phone, then the transmitting module 24 uses the mobile video stream signal (3gP, _)' to actually transmit the location video to the client device 12. If the client device 12 is a computer, the transmission module can transmit the file in a format such as WMA. Please refer to the 3A side, and the 3A to 3B are the schematic diagrams of the orientation of the hair-control camera. As shown in Fig. 3B, the intelligent monitoring and analysis system 1 of the present invention mainly includes the packet 14, the intrusion judging module 16, the _ tracking module ΐδ, and the image database 2. , the clear face module 22, the sending module 24, and the format conversion module sink. / In order to make it possible for the user to be able to use the user's position, the user can pass the operation as shown in Figure 3A: the brother has more to do the game. _ New, Money 岐 through the display interface 1 1338504 ', the operation module 36 will move the camera 14 to the orientation information relative to the orientation to the right as shown in Figure 3b, and then convey the orientation information to the camera 14 , the location of the news to shoot the location. In addition, when the intrusion determination module 16 provides the object outline of the invading object, the detection tracking module 18 moves the upper, lower, left and right when the object frame in the opposite position contains the invading object. The camera 14 automatically shoots at the center point of the invading object to obtain the most complete face. 2 Μ Μ 4A~4B 图 图 Please refer to section 4A, which is the intention of the shooting of the heart. As shown in FIG. 4A, the intrusion determination module 16 can divide the received location frame into nine control cells (the control cell in the center is the control cell 4, for example, the right input object only spans When the control points 40a, Yang, and the center point of the invading object only fall on the control grid, the camera 14 is automatically moved to the right so that the center point of the invading age falls in the control grid. If the center of the invading object When the spring is already in control, the control signal will not be sent to the camera u, and the camera 14 will remain intact. Thus, the camera 14 can automatically continue to track the invading objects and keep the objects in the surveillance camera. In the monitoring surface of the cymbal, until the intrusion object movement monitoring camera 14 can monitor the range of β, by the above-mentioned better specific implementation of the Wei, the scale can more clearly describe the characteristics and spirit of the present invention. The disclosure is to be construed as limiting the scope of the invention. The object of the invention is intended to cover various modifications and equivalents. 16 1338504 [Simple description of the drawings] 1A to 1B are schematic diagrams for transmitting a clear picture according to the present invention. 2A to 2B are schematic diagrams of video files of a transmission place according to the present invention. 3A to 3B are schematic diagrams showing the orientation of a remote control camera of the present invention. 4A to 4B are schematic diagrams of the center point shooting of the present invention. [Main component symbol description] 10 intelligent monitoring and analyzing system 12 user terminal device 14 camera 16 intrusion judging module 18 detecting tracking module 20 image database • 22 Clear picture pole group 24 transmission module 26 format conversion module 30 display 30a~30d block 32 operation interface 34 display module 36 operation module 17 1338504 40a~40b control grid

18

Claims (1)

1338504 X. Patent application scope: A smart monitoring and analysis tool, such as the monitoring system, is used to monitor the intelligent monitoring and analysis system including: ===:), get _ photo,,, from I small / 71 T人Tj卿丨棋 An object-containing object that is invaded and removed from the frame of the opposite material. The intrusive tracking module is used to provide the object of the invading object in the intrusion determination module. When the contour is in the opposite position, the intrusion object is included in the map frame, and the pen field frames containing the invading object are initially stored as a local video file; A clear facet module is configured to find the place frame of the object frame that has the largest proportion of the place frame from the pen place frame containing the invading object, and define the place frame frame Forming a clear face; and a sending module 'for transmitting the clear face, the resource address (Uniform Resource Locator, URL) of the place sfl file and/or the location video file itself to 19 a client device . 2 As for the intelligent monitoring and analysis department described in item 1 of the scope of the patent application, '’ /, medium. When the intrusion object is included in the location frame, the black detection tracking module starts to store the pen field frames related to the intrusion object into the location video file until the location frame no longer contains the Invasive objects. 3. The intelligent monitoring and analysis system as described in claim 1 of the patent application, wherein the format of the visual file of the location is relative to the format that the client device can display. 4. The intelligent monitoring and analysis system as described in the scope of the patent application, wherein the client device can be a mobile phone supporting a GPRS*3G network or a computer device supporting the Internet and a browser. < 5, the intelligent monitoring and analysis system of claim 1, wherein the client device further comprises: a display module for presenting the clear picture or/and the location video through a display An operation module is configured to provide an operation interface for receiving a position-orientation signal that the user refers to, and moving the camera to a position relative to the orientation information to capture the location. 20 1338504 6. The intelligent monitoring and analysis system according to claim 5, wherein the client device receives the clear address transmitted from the sending module, and the resource address of the video building in the location. (Uniform Resource Locator, URL) and/or after the video file itself, when the user requests to play the video file of the location, the display module transmits the location video in a streaming packet format according to the resource address of the video file of the location. File and display the location video file through the display. 7. The intelligent monitoring and analysis system according to claim 1, wherein the detection and tracking module determines that the relative location of the object is included in the intrusion determination module to provide the object contour of the intrusion object. When the object is invaded, the camera is aimed at the nail point of the invading object. 8. The intelligent monitoring and analysis system of claim i, wherein the clear picture module is used to find the outline of the object from the pen-like frame containing the invading object (4) The opposite part of the face of the face, and the relative frame of the % is defined as a clear face. 9, a kind of intelligent monitoring and analysis method, used to monitor a place, the intelligent monitoring analysis method includes: 21 1338504 I::: and obtain a continuous shooting of multiple pen maps from the pen place frame The inclusion-intrusion object; "object: the outline of the object of the invading object in the frame of the sea; the relative position of the site is determined to be open when the object is opened. The pen place frames are stored as a place video file; k containing the invading object in the frame of the pico ^ ^ 2聿琢, to find out the most frame of the proportion of the frame occupied by the force, and This will be defined as a clear written copy of the frame of the person, and the source of the video of the location of the video. Nif〇rm Resource Locator, URL)^/^ The intelligent monitoring and analysis party described in the file itself to a user terminal, such as ^(4)_9, will start the video related to the intrusion when the intrusion object is contained in the location box (4). ^Place frame is stored as the invasion The object of the intelligent monitoring and analysis method mentioned in item 9 of the patent application scope is no longer included in the frame of the sea. The clear picture of the video file of the place is 10 The format that can be displayed. For example, the smart monitoring and analysis party described in claim 9 can be a mobile phone supporting GPRS or 3G network, or a computer device supporting the Internet and a browser. The intelligent monitoring and analysis method according to claim 9, wherein the client device further comprises: a display module for presenting the clear face or/and the location video slot through a display The operation group is configured to provide an operation interface for receiving a position information indicated by the user, and moving the camera to a position relative to the position information to photograph the location. 14. The intelligent monitoring and analysis method according to claim 13, wherein the client device receives the clear picture transmitted from the sending module, and the resource address of the video file of the place (Uniform Resource) After L〇cat〇r, url^〇/ or the video file itself of the location, when the user requests to play the video file of the location, the display module will use the resource address of the location video 23 to solve the case. And through the (four) 'transport handle X'., the staff does not present the video file of the place. 15\. If the application of the patent model, the smart monitoring and analysis party to the eight to determine the relative location map When the frame contains the intrusion object, the camera is aimed at the center point of the invading object. 16. The intelligent monitoring and analysis method according to claim 9 of the patent application, further comprising: the In the pen map frame, find the object frame with the identifiable face of the object, and define the relative frame as a clear face.