WO2017163282A1 - 監視装置及び監視システム - Google Patents
監視装置及び監視システム Download PDFInfo
- Publication number
- WO2017163282A1 WO2017163282A1 PCT/JP2016/004148 JP2016004148W WO2017163282A1 WO 2017163282 A1 WO2017163282 A1 WO 2017163282A1 JP 2016004148 W JP2016004148 W JP 2016004148W WO 2017163282 A1 WO2017163282 A1 WO 2017163282A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- camera
- time
- person
- monitoring
- video
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
- G08B13/19602—Image analysis to detect motion of the intruder, e.g. by frame subtraction
- G08B13/19608—Tracking movement of a target, e.g. by detecting an object predefined as a target, using target direction and or velocity to predict its new position
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0007—Image acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/254—Analysis of motion involving subtraction of images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
- G08B13/19639—Details of the system layout
- G08B13/19645—Multiple cameras, each having view on one of a plurality of scenes, e.g. multiple cameras for multi-room surveillance or for tracking an object by view hand-over
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
Definitions
- the present disclosure relates to a monitoring device and a monitoring system for identifying a person reflected in a camera and tracking the identified person.
- Patent Document 1 discloses a monitoring system including a plurality of monitoring cameras.
- the surveillance camera extracts feature information of the object shown in the video and transmits the feature information to other surveillance cameras.
- a plurality of monitoring cameras can cooperate to track and monitor an object having the same feature information.
- This disclosure provides a monitoring device and a monitoring system effective for accurately tracking an object.
- a monitoring apparatus is a monitoring apparatus that identifies an object from videos captured by a plurality of cameras having a predetermined positional relationship including a first camera and a second camera, and receives videos from the plurality of cameras.
- a storage unit that stores feature information that represents the feature of the object, camera placement information that represents the placement position of the camera, and a control unit that identifies the object from the video based on the feature information.
- the control unit can identify an object that can be identified from the video captured by the first camera but cannot be identified from the video captured by the second camera, based on the camera arrangement information. It is specified in the shot video.
- the monitoring device and the monitoring system according to the present disclosure are effective for accurately tracking an object.
- FIG. 1 is a block diagram showing a configuration of a monitoring system according to a first embodiment.
- 7 is a flowchart for explaining an operation of calculating a moving time between a person identification and a monitoring camera in the first embodiment.
- (A) is a figure for demonstrating the feature extraction of a person
- (b) is a figure which shows an example of the feature information table in Embodiment 1.
- FIG. FIG. 5 is a diagram illustrating an example of a shooting time information table according to the first embodiment.
- FIG. 6 is a diagram illustrating an example of a camera arrangement information table according to the first embodiment. 6 is a flowchart for explaining an operation for correcting a shooting time information table according to the first embodiment.
- FIG. The figure for demonstrating scoring of distribution of the number of people who move between the some surveillance cameras in other embodiment
- Embodiment 1 will be described with reference to the drawings. In this embodiment, even if a situation occurs in which a feature of an object cannot be extracted from a part of a plurality of monitoring cameras, a monitoring system effective for tracking the object is provided.
- FIG. 1 shows the configuration of the monitoring system of the first embodiment.
- the monitoring system 100 uses a plurality of monitoring cameras 1 (monitoring cameras a, b, c, and d) and images captured by the plurality of monitoring cameras 1, and identifies a person shown in the images. And a monitoring device 2 for identifying and tracking.
- monitoring cameras 1 monitoring cameras a, b, c, and d
- images captured by the plurality of monitoring cameras 1 and identifies a person shown in the images.
- a monitoring device 2 for identifying and tracking.
- Each monitoring camera 1 includes a photographing unit 11 that photographs a video, and a transmission unit 12 that transmits the video photographed by the photographing unit 11 to the monitoring device 2.
- the photographing unit 11 can be realized by a CCD image sensor, a CMOS image sensor, an NMOS image sensor, or the like.
- the transmission unit 12 includes an interface circuit for performing communication with an external device in compliance with a predetermined communication standard (for example, LAN, WiFi).
- the monitoring device 2 includes a receiving unit 21 that receives the video from each monitoring camera 1, a video storage unit 22a that stores the received video, and an object shown in the video stored in the video storage unit 22a (this embodiment) And a control unit 23 that identifies the person) and tracks the identified object.
- the receiving unit 21 includes an interface circuit for performing communication with an external device in accordance with a predetermined communication standard (for example, LAN, WiFi).
- the control unit 23 can be realized by a semiconductor element or the like.
- the function of the control unit 23 may be configured only by hardware, or may be realized by combining hardware and software.
- the control unit 23 can be configured by, for example, a microcomputer, CPU, MPU, DSP, FPGA, and ASIC.
- the control unit 23 includes a recognition unit 23a for identifying an object shown in the video stored in the video storage unit 22a.
- the recognizing unit 23a extracts the feature of the object shown in the video stored in the video storage unit 22a, generates feature information representing the feature, and the target having the extracted feature is sent to the monitoring camera 1.
- Shooting time information representing the time zone being shown is generated.
- the feature information and the shooting time information are recognition information obtained by recognizing an object.
- the monitoring device 2 further includes a recognition information storage unit 22b that stores a feature information table T1 and a shooting time information table T2, and a camera placement information storage unit 22c that stores a camera placement information table T3.
- the feature information table T1 includes feature information of the object generated by the recognition unit 23a.
- the shooting time information table T2 includes shooting time information generated by the recognition unit 23a.
- the camera arrangement information table T3 includes information representing the arrangement position of the monitoring camera 1 and the time taken for the object to move between the monitoring cameras.
- the control unit 23 further calculates a time required for the movement between the monitoring cameras based on the photographing time information table T2, and updates the camera arrangement information table T3.
- the feature information table T1 A recognition information correction unit 23c that corrects the shooting time information table T2 based on the camera arrangement information table T3.
- the recognition information correction unit 23c identifies the monitoring camera 1 where the object is to be reflected, determines whether or not the object is reflected in the identified monitoring camera 1, and displays the identified image.
- the time zone in which the subject that is not reflected is supposed to appear in the surveillance camera 1 is calculated (estimated).
- one of the object candidates displayed on the monitoring camera 1 in the calculated (estimated) time zone is specified as the object determined not to be displayed, and the photographing time information table T2 is corrected.
- the video storage unit 22a, the recognition information storage unit 22b, and the camera arrangement information storage unit 22c are the same or separate storage units that can be realized by, for example, a DRAM, a ferroelectric memory, a flash memory, or a magnetic disk.
- the monitoring device 2 further has a display unit 24.
- the display unit 24 can display the video stored in the video storage unit 22a, the feature information table T1, and the shooting time information table T2.
- the display unit 23 can be realized by a liquid crystal display or the like.
- FIG. 2 the example of arrangement
- positioning of the surveillance camera 1 is shown.
- the surveillance camera 1 is provided in a store, for example.
- the four surveillance cameras 1 (surveillance cameras a, b, c, d) are arranged in different places.
- the monitoring cameras a, b, and c are arranged in the order of the traveling direction (the direction from the left side to the right side in FIG. 2).
- the number of surveillance cameras 1 and their locations are merely examples, and can be arbitrarily changed.
- Each of the monitoring cameras 1 transmits the video imaged by the imaging unit 11 from the transmission unit 12 to the monitoring device 2.
- the transmitted video is stored in the video storage unit 22a of the monitoring device 2.
- FIG. 3 shows a process of identifying a person and calculating a moving distance between the monitoring cameras by the control unit 23.
- the control unit 23 performs the person identification process shown in FIG. 3 at a predetermined timing.
- the predetermined timing may be when the user instructs the monitoring device 2 or may be every predetermined time (for example, 24 hours).
- the monitoring camera 1 captures a person who has moved in the traveling direction indicated by the arrow in FIG. 2 will be described as an example.
- the recognition unit 23a reads the video stored in the video storage unit 22a, and extracts the characteristics of the person shown in the video (S301). For example, the recognizing unit 23a sequentially analyzes the video from the video of the monitoring camera a. The recognition unit 23a extracts, for example, the shape, color, size, or position of a part of the face as the feature of the person.
- FIG. 4A shows an example of person feature extraction
- FIG. 4B shows an example of the feature information table T1.
- the recognizing unit 23a as the characteristics of a person, for example, as shown in FIG. 4A, the distance between both eyes (distance between “I-II”) and the distance between one eye and the nose (“II-III”). ”Is extracted, and feature information 41 including the extracted feature (distance) is added to the feature information table T1 as shown in FIG.
- the recognizing unit 23a determines whether or not the feature information 41 indicating the feature that matches the extracted feature already exists in the feature information table T1 (S302).
- the identification information (ID) for identifying the person is generated, and the generated identification information and the characteristics of the person (the distance between “I-II” and “II-III” are generated.
- the feature information 41 including the distance between them is added to the feature information table T1 (S303).
- the recognizing unit 23a generates shooting time information indicating when the person is shown on which monitoring camera 1, and adds the shooting time information to the shooting time information table T2 (S304).
- FIG. 5 shows an example of the shooting time information table T2.
- the shooting time information 51 includes the identification information (ID) of the person, the identification information of the monitoring camera 1 that shot the person, the time when the person started to appear on the monitoring camera 1 (IN time), and the time when the reflection ends (OUT time). ).
- the recognizing unit 23a determines whether or not the reading of the images from all the monitoring cameras 1 has been completed (S305). If the reading has not been completed, the processing of steps S301 to S304 is performed for the remaining monitoring camera 1 images. repeat.
- the moving time information updating unit 23b updates the camera arrangement information table T3 based on the shooting time information table T2 generated by the recognizing unit 23a.
- FIG. 6 shows an example of the camera arrangement information table T3.
- the camera arrangement information table T3 includes arrangement information 61 indicating the arrangement position of the monitoring camera 1 and movement time information 62 indicating the time taken for movement between the monitoring cameras 1.
- the arrangement information 61 is identification information of “current camera (current)” and “next camera (next)” that appears when a person travels along the movement path (in the traveling direction indicated by the arrow in FIG. 2). including.
- the arrangement information 61 is input in advance.
- the movement time information 62 includes information on the shortest time and the longest time required for movement from the “current camera” to the “next camera”.
- the movement time information update unit 23b updates the shortest time and the longest time in the camera arrangement information table T3 based on the IN time and the OUT time in the photographing time information table T2.
- the travel time information updating unit 23b may update the travel time information 62 in the camera arrangement information table T3 periodically without being limited to the timing of step S306.
- the person can be identified by extracting the characteristics of the person from the video photographed by the monitoring camera 1, and when the person identified by referring to the photographing time information table T2 appears in which monitoring camera 1. You can recognize what happened. Accordingly, it is possible to track a person using the images of the plurality of monitoring cameras 1.
- monitoring cameras 1 for example, monitoring cameras a, b, c
- the person from some monitoring cameras 1 cannot be extracted, and the tracking of the person may be interrupted.
- the same person is extracted using the camera arrangement information table T3, and the photographing time information table T2 is corrected.
- FIG. 7 shows a process of correcting the photographing time information table T2 by the recognition information correcting unit 23c.
- the recognition information correction unit 23c reads out the photographing time information table T2 and rearranges them in the descending order of the number of person entries (S701).
- FIG. 8 shows the photographing time information table T2 after rearrangement.
- the recognition information correction unit 23c extracts one person in descending order of the number of entries (S702), and refers to the camera arrangement information table T3 to check whether or not the shooting time information 51 is missing for the extracted person. (S703). Whether or not the shooting time information 51 is missing is confirmed by referring to the arrangement information 61 in the camera arrangement information table T3 and specifying the monitoring camera 1 where a person should appear.
- the recognition information correction unit 23c can determine that it should be reflected on the monitoring camera b before being displayed on the monitoring camera c. That is, when there is the shooting time information 51 of the monitoring camera c, the monitoring camera b is specified as the monitoring camera 1 that should appear, and the presence or absence of the shooting time information 51 of the monitoring camera b is confirmed. In this way, it is determined whether or not the shooting time information 51 is missing.
- the recognition information correcting unit 23c refers to the photographing time information table T2 and the camera arrangement information table T3 and is reflected in the monitoring camera 1 in which the photographing time information 51 is missing.
- An estimated time zone is estimated (calculated) (S705). For example, as shown in FIG. 8, for the person B, the shooting time information 51 of the monitoring camera b is missing.
- the recognition information correction unit 23c outputs the OUT time (10:19) of the monitoring camera a of the person B in the shooting time information table T2 and the shortest time required for the movement from the monitoring camera a to the monitoring camera b in the camera arrangement information table T3.
- a time zone (10:29 to 10:31) at which the person B starts to appear on the monitoring camera b is estimated.
- the recognition information correcting unit 23c is related to the IN time (10:41) of the monitoring camera c of the person B in the shooting time information table T2 and the movement from the monitoring camera b to the monitoring camera c in the camera arrangement information table T3. From the shortest time (5 minutes) and the longest time (6 minutes), the time period (10:35 to 10:36) at which the person B finishes appearing on the surveillance camera b is estimated.
- the recognition information correction unit 23c extracts a person appearing in the estimated time zone from the shooting time information table T2 (S706).
- the IN time (10:31) of the monitoring camera b of the person E is included in the estimated start time zone (between 10:29 and 10:31) and Since the OUT time (10:36) of the monitoring camera b is included in the estimated end time zone (between 10:35 and 10:36), the recognition information correction unit 23c extracts the person E. .
- the recognition information correction unit 23c extracts the shooting time information 51 from the missing persons. If the extracted person is one, it is determined that the person (person E) is the same person (person B), and the photographing time information table T2 is corrected.
- the identification information of the person B is recorded in “corrected person identification information (ID)” for the person E.
- the recognition information correcting unit 23c determines that the person with the closest feature information 41 is the same person based on the feature information table T1, and the shooting time information table T2 is corrected (S707).
- the recognition information correcting unit 23c displays the shooting time information table T2 on the display unit 24 ( S709). The user can confirm the tracking of the person photographed by the plurality of monitoring cameras 1 by referring to the corrected photographing time information table T2 displayed on the display unit 24.
- the recognition information correction unit 23c compensates for the lack of the shooting time information 51 by using the feature information table T1, the shooting time information table T2, and the camera arrangement information table T3.
- the recognition information correction unit 23c corrects the shooting time information 51 of the monitoring camera b for the person B by correcting the person E as the person B as shown in FIG.
- the control unit 23 can track the person B from the images shot in the order of the monitoring camera a, the monitoring camera b, and the monitoring camera c with reference to the corrected shooting time information table T2. Become.
- the monitoring device 2 of the present embodiment was photographed by the plurality of monitoring cameras 1 including the first camera (monitoring camera c) and the second camera (monitoring camera b) and having a predetermined positional relationship.
- a monitoring device for identifying an object from video a receiving unit 21 for receiving video from a plurality of monitoring cameras 1, a recognition information storage unit 22b for storing characteristic information 41 representing characteristics of the target, and a camera arrangement
- a camera arrangement information storage unit 22c that stores arrangement information 61 representing a position; and a control unit 23 that identifies an object from the video based on the feature information 41.
- the recognition information correction unit 23c of the control unit 23 includes Based on the arrangement information 61, an object that could be identified from the video captured by the first camera (monitoring camera c) but could not be identified from the video captured by the second camera (monitoring camera b) Second mosquito Identified in image photographed by the La (surveillance cameras b). As described above, since the object that cannot be identified by the feature information 41 is specified using the arrangement information 61, the object can be accurately tracked.
- the movement time information update unit 23b of the control unit 23 calculates the movement time of the person between the first camera and the second camera, the calculated movement time, and the time when the object is photographed by the first camera. Based on the above, the time zone in which the object passes through the shooting area of the second camera is calculated, and the object is specified in the video shot by the second camera in the calculated time zone. Specifically, shooting time information 51 representing a time zone in which each object identified based on the feature information 41 is reflected in the monitoring camera 1 is generated, and between the monitoring cameras 1 based on the generated shooting time information 51 Calculate the time taken to move.
- control unit 23 identifies the monitoring camera 1 on which each object should appear based on the arrangement information 61, and when the object is not reflected on the identified monitoring camera 1, the calculated movement between the monitoring cameras Based on the time it takes, the time zone in which the object that is not shown is supposed to appear in the specified surveillance camera 1 is estimated, and the shooting time information 51 is referred to, and the subject is reflected in the specified surveillance camera 1 in the estimated time zone. It identifies that the other target object is not reflected, and rewrites the shooting time information 51. Thus, even if the feature information 41 does not match and the shooting time information 51 is missing, the missing information is obtained by referring to the camera placement information table T3 including the placement information 61 and the travel time information 62.
- the generated shooting time information 51 can be supplemented. Therefore, the feature information 41 of the object (person) acquired from the video of some monitoring cameras 1 is not acquired from the video of other monitoring cameras 1, but is different from that of other monitoring cameras 1. Even when it is recognized as an object (person), it can be re-recognized as the same object (person) by referring to the camera arrangement information table T3. Therefore, it is possible to accurately track the object (person).
- the control unit 23 selects two or more target object candidates based on the feature information 41. One of them is identified as an object. Thereby, even when two or more target object candidates are shown in the video, it is possible to accurately specify the target object determined not to be shown from the target object candidates.
- control unit 23 extracts the feature of the object from the video received by the receiving unit 21, generates the feature information 41, and stores it in the recognition information storage unit 22b. Thereby, even when a feature of a target object is newly extracted, the target object can be identified and tracked.
- the monitoring system 100 represents a plurality of monitoring cameras 1 including a first camera and a second camera and having a predetermined positional relationship, feature information 41 representing features of an object, and an arrangement position of the monitoring camera 1.
- a second camera that has an arrangement information 61, and based on the feature information 41, the object can be identified from the images photographed by the plurality of surveillance cameras 1 and can be identified from the images photographed by the first camera.
- a monitoring device 2 that identifies an object that could not be identified from the video imaged in (2) based on the arrangement information 61 in the video imaged by the second camera.
- an object person
- the monitoring system 100 is also useful for simulating a change in the flow line and analyzing the value of the store area.
- the first embodiment has been described as an example of the technique disclosed in the present application.
- the technology in the present disclosure is not limited to this, and can also be applied to embodiments in which changes, replacements, additions, omissions, and the like have been made as appropriate.
- the recognition information correction unit 23c refers to the shooting time information table T2 and the camera arrangement information table T3, and the lower time of the time zone in which the person B starts to appear on the monitoring camera b (10:29)
- the person shown in the surveillance camera b is extracted in any time zone (for example, 10:32 to 10:35) from the time until the upper limit time (10:36) (S706).
- a first predetermined time for example, 3 minutes is further determined from the lower limit time (10:29) of the estimated start time zone of the projection.
- the second predetermined time (for example, 3 minutes) is added to the upper limit time (10:36) of the estimated end time of the projection, and a new time zone (10:26 to 10:39) is subtracted. Minutes) and is reflected in the monitoring camera b in any time zone (for example, 10: 28-10: 34) in the new time zone (10: 26-10: 39) A person may be extracted.
- the most similar person may be extracted from a plurality of persons according to the sum of the similarity degree based on the feature information 41 and the appearance degree based on the probability distribution of the time required for movement between the monitoring cameras.
- a case will be described below in which the person A cannot be detected from the video of the monitoring camera b in FIG. 2 and the person A is searched from the persons B, C, and D shown on the monitoring camera b.
- the similarity S f (A, x) based on the feature information and the appearances S ab (t 1 ), S bc (t 2 ) based on the probability distribution of the time required for movement between the monitoring cameras are functionalized.
- the person whose total value (total score) S (A, x) represented by the following formula (1) satisfies the predetermined condition is the person A.
- the person with the highest total value S (A, x) is determined as person A.
- S (A, x) S f (A, x) + ⁇ S ab (t 1 ) + ⁇ S bc (t 2 )
- X is person B, C, D
- S f (A, x) is the similarity between the feature information of person A and the feature information of person x)
- ⁇ and ⁇ are predetermined weighting coefficients
- T 1 is the time from when the person A appears on the surveillance camera a until the person x appears on the surveillance camera b
- T 2 is the time from appearing in person x surveillance cameras b, to the person A is reflected in surveillance camera c
- S ab (t) is the degree of appearance based on the time taken to move from the surveillance camera a to the surveillance camera b and the appearance frequency distribution of the person
- S bc (t) is the degree of appearance based on the time taken to move from the surveillance camera b to the surveillance camera c and the appearance frequency distribution of the person)
- FIG. 9A shows an example of the appearance degree S ab (t).
- the degree of appearance S ab (t) is determined for each time period from when the person no longer appears on the monitoring camera a until the person starts to appear on the monitoring camera b (ie, the time taken to move from the monitoring camera a to the monitoring camera b). This is a function of the distribution of appearance frequencies.
- FIG. 9B shows an example of the appearance degree S bc (t).
- the degree of appearance S bc (t) is determined for each time period from when the person stops appearing on the monitoring camera b until the person starts to appear on the monitoring camera c (that is, the time taken to move from the monitoring camera b to the monitoring camera c). This is a function of the distribution of appearance frequencies.
- the recognition information correction unit 23c refers to the shooting time information table T2, calculates the appearance frequency of the person for each time required for movement between the monitoring cameras, and calculates the time required for the movement between the monitoring cameras and the calculated appearance frequency of the person. Based on the above, the function S ab (t) and S bc (t) of the appearance degree as shown in FIG. 9A and FIG. 9B are generated.
- FIG. 10 shows an example in which the numerical value of the total value S (A, x) represented by the above formula (1) is obtained by the numerical values of S f (A, x), t 1 , and t 2 , respectively.
- the total value S (A, x) of the person D is the highest, the person D is determined to be the person A.
- the functions S ab (t) and S bc (t) may change depending on the person, the time zone, the store situation, and the like.
- the functions S ab (t) and S bc (t) may be generated for each time zone based on the current time (9:00 to 10:00, etc.).
- the recognition information correction unit 23c determines that the person (person B) from which the shooting time information 51 is missing is estimated.
- the feature information 41 with the person (person E) shown in the band may be compared. If the feature information 41 is not similar, the recognition information correction unit 23c determines that the person (person E) who is reflected in the estimated time zone is the person (person B) from which the shooting time information 51 is missing. It is not necessary to correct the photographing time information table T2 by determining that the person is another person.
- the monitoring system 100 of the present disclosure can be realized by cooperating with hardware resources such as a processor, a memory, and a program.
- the present disclosure is applicable to a monitoring apparatus that tracks a target object using a plurality of monitoring cameras and a monitoring system having the monitoring apparatus.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Closed-Circuit Television Systems (AREA)
- Image Analysis (AREA)
- Alarm Systems (AREA)
- Burglar Alarm Systems (AREA)
Abstract
Description
実施形態1について、図面を用いて説明する。本実施形態において、複数の監視カメラのうちの一部において対象物の特徴を抽出できない状況が生じても、その対象物を追跡するのに有効な監視システムを提供する。
図1は、実施形態1の監視システムの構成を示している。本実施形態の監視システム100は、複数の監視カメラ1(監視カメラa,b,c,d)と、複数の監視カメラ1で撮影された映像を使用して、その映像に映っている人物を識別して追跡する監視装置2とを含む。
図2に、監視カメラ1の配置例を示す。監視カメラ1は、例えば、店舗に設けられる。4台の監視カメラ1(監視カメラa,b,c,d)は、それぞれ異なる場所に配置される。図2においては、監視カメラa,b,cが進行方向(図2の左側から右側の方向)の順に配置されている。なお、監視カメラ1の数とその配置箇所は単なる例であって、任意に変更可能である。監視カメラ1はそれぞれ、撮影部11が撮影した映像を、送信部12から監視装置2に送信する。送信された映像は、監視装置2の映像蓄積部22aに格納される。
撮影したときの角度や照明条件によって同一人物であっても映像への映り方が異なることがある。そのため、複数の監視カメラ1で撮影した映像から抽出される同一人物の特徴が一致しないことがある。たとえば、明るい場所で高い位置に設置された監視カメラ1と、暗い場所で低い位置に設置された監視カメラ1では、撮影される映像が大きく異なるため、両者で撮影される映像から抽出される人物の特徴は異なる場合がある。この場合、同一人物であっても、抽出した特徴が異なるために、他の人物として認識してしまう。そのため、同一人物が複数の監視カメラ1(例えば、監視カメラa,b,c)の前を順に通った場合であっても、一部の監視カメラ1(例えば、監視カメラb)からはその人物の特徴を抽出できず、その人物の追跡が途切れてしまうことがある。
以上のように、本実施形態の監視装置2は、第一のカメラ(監視カメラc)及び第二のカメラ(監視カメラb)を含み所定の位置関係を有する複数の監視カメラ1によって撮影された映像から対象物を識別する監視装置であって、複数の監視カメラ1から映像を受信する受信部21と、対象物の特徴を表す特徴情報41を格納する認識情報蓄積部22bと、カメラの配置位置を表す配置情報61を格納するカメラ配置情報蓄積部22cと、特徴情報41に基づいて映像から対象物を識別する制御部23と、を備え、制御部23の認識情報修正部23cは、第一のカメラ(監視カメラc)で撮影された映像から識別できたが、第二のカメラ(監視カメラb)で撮影された映像から識別できなかった対象物を、配置情報61に基づいて、第二のカメラ(監視カメラb)で撮影された映像において特定する。このように、特徴情報41で識別できなかった対象物を、配置情報61を使用して特定するため、対象物の追跡を精度良く実現できる。
以上のように、本出願において開示する技術の例示として、実施形態1を説明した。しかしながら、本開示における技術は、これに限定されず、適宜、変更、置き換え、付加、省略などを行った実施形態にも適用可能である。また、上記実施形態1で説明した各構成要素を組み合わせて、新たな実施形態とすることも可能である。そこで、以下、他の実施形態を例示する。
S(A,x)=Sf(A,x)+αSab(t1)+βSbc(t2)・・・(1)
(xは、人物B,C,D)
(Sf(A,x)は、人物Aの特徴情報と人物xの特徴情報の類似度)
(α、βは、所定の重み付けの係数)
(t1は、人物Aが監視カメラaに映ってから、人物xが監視カメラbに映るまでの時間)
(t2は、人物xが監視カメラbに映ってから、人物Aが監視カメラcに映るまでの時間)
(Sab(t)は、監視カメラaから監視カメラbへの移動にかかる時間と人物の出現頻度の分布に基づく出現度)
(Sbc(t)は、監視カメラbから監視カメラcへの移動にかかる時間と人物の出現頻度の分布に基づく出現度)
2 監視装置
11 撮影部
12 送信部
21 受信部
22a 映像蓄積部
22b 認識情報蓄積部
22c カメラ配置情報蓄積部
23 制御部
23a 認識部
23b 移動時間情報更新部
23c 認識情報修正部
24 表示部
100 監視システム
T1 特徴情報テーブル
T2 撮影時刻情報テーブル
T3 カメラ配置情報テーブル
Claims (9)
- 第一及び第二のカメラを含み所定の位置関係を有する複数のカメラによって撮影された映像から対象物を識別する監視装置であって、
前記複数のカメラから映像を受信する受信部と、
前記対象物の特徴を表す特徴情報と、前記カメラの配置位置を表すカメラ配置情報とを格納する記憶部と、
前記特徴情報に基づいて前記映像から前記対象物を識別する制御部と、
を備え、
前記制御部は、前記第一のカメラで撮影された映像から識別できたが、前記第二のカメラで撮影された映像から識別できなかった前記対象物を、前記カメラ配置情報に基づいて、前記第二のカメラで撮影された映像において特定する、
監視装置。 - 前記制御部は、
前記第一のカメラと前記第二のカメラ間の前記対象物の移動時間を算出し、
前記算出した移動時間と、前記第一のカメラにおいて前記対象物が撮影された時間とに基づき、前記第二のカメラの撮影領域を前記対象物が通過した時間帯を算出し、
前記算出した時間帯に前記第二のカメラにより撮影された映像において前記対象物を特定する、請求項1に記載の監視装置。 - 前記制御部は、前記算出した時間帯に前記第二のカメラにより撮影された映像において2つ以上の対象物候補が映っている場合、前記特徴情報に基づいて、前記2つ以上の対象物候補の中から一つを前記対象物として特定する、請求項2に記載の監視装置。
- 前記制御部は、前記受信部が受信した映像から前記対象物の特徴を抽出して前記特徴情報を生成して、前記記憶部に格納する、請求項1に記載の監視装置。
- 前記第一のカメラと前記第二のカメラは、前記対象物の移動経路に沿った撮影領域を撮影する位置に配置される、請求項1に記載の監視装置。
- 前記第一のカメラと前記第二のカメラは、前記対象物の移動経路に沿った撮影領域を撮影する位置に配置され、
前記配置が、前記対象物が前記第一のカメラの撮影領域の次に前記第二のカメラの撮影領域を通過する配置の場合、前記制御部は、前記第一のカメラの撮影終了時刻に、前記算出した移動時間を加算して、前記第二のカメラの撮影領域を前記対象物が通過した時間帯の開始時刻を算出し、
前記配置が、前記対象物が前記第二のカメラの撮影領域の次に前記第一のカメラの撮影領域を通過する配置の場合、前記制御部は、前記第一のカメラの撮影開始時刻から、前記算出した移動時間を減算して、前記第二のカメラの撮影領域を前記対象物が通過した時間帯の終了時刻を算出する、請求項2に記載の監視装置。 - 前記算出した時間帯に、前記第二のカメラにより撮影された映像において対象物候補が映っていない場合、前記制御部は、前記開始時刻から第1の所定時間を減算し、又は前記終了時刻に第2の所定時間を加算することによって、新たな時間帯を算出する、請求項6に記載の監視装置。
- 前記制御部は、前記算出した時間帯に前記第二のカメラにより撮影された映像において2つ以上の対象物候補が映っている場合、前記第一のカメラと前記第二のカメラ間の移動時間の確率分布に基づいて、前記2つ以上の対象物候補の中から一つを前記対象物として特定する、請求項2に記載の監視装置。
- 第一及び第二のカメラを含み所定の位置関係を有する複数のカメラと、
対象物の特徴を表す特徴情報と、前記カメラの配置位置を表すカメラ配置情報とを有し、前記特徴情報に基づいて、前記複数のカメラによって撮影された映像から前記対象物を識別し、前記第一のカメラで撮影された映像から識別できたが、前記第二のカメラで撮影された映像から識別できなかった前記対象物を、前記カメラ配置情報に基づいて、前記第二のカメラで撮影された映像において特定する、請求項1から請求項8のいずれかに記載の監視装置と、
を含む、監視システム。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2018506503A JP6688975B2 (ja) | 2016-03-25 | 2016-09-12 | 監視装置及び監視システム |
EP16895322.2A EP3435665A4 (en) | 2016-03-25 | 2016-09-12 | MONITORING AND MONITORING SYSTEM |
US16/139,527 US10863142B2 (en) | 2016-03-25 | 2018-09-24 | Monitoring device and monitoring system |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2016062031 | 2016-03-25 | ||
JP2016-062031 | 2016-03-25 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US16/139,527 Continuation US10863142B2 (en) | 2016-03-25 | 2018-09-24 | Monitoring device and monitoring system |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2017163282A1 true WO2017163282A1 (ja) | 2017-09-28 |
Family
ID=59899998
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2016/004148 WO2017163282A1 (ja) | 2016-03-25 | 2016-09-12 | 監視装置及び監視システム |
Country Status (4)
Country | Link |
---|---|
US (1) | US10863142B2 (ja) |
EP (1) | EP3435665A4 (ja) |
JP (1) | JP6688975B2 (ja) |
WO (1) | WO2017163282A1 (ja) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110049290A (zh) * | 2019-04-24 | 2019-07-23 | 北京市亚太安设备安装有限责任公司 | 公共建筑物智能监控系统 |
CN112689132A (zh) * | 2021-03-15 | 2021-04-20 | 成都点泽智能科技有限公司 | 目标对象监控方法和监控设备 |
JP2021093052A (ja) * | 2019-12-12 | 2021-06-17 | 清水建設株式会社 | 情報修正装置、情報修正方法およびプログラム |
US11250251B2 (en) * | 2018-09-06 | 2022-02-15 | Nec Corporation | Method for identifying potential associates of at least one target person, and an identification device |
JP2022051683A (ja) * | 2020-09-22 | 2022-04-01 | グラスパー テクノロジーズ エーピーエス | 訓練データの生成と再識別に使用するための機械学習モデルの訓練とについての概念 |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11501568B2 (en) * | 2018-03-23 | 2022-11-15 | Nec Corporation | Information processing apparatus, person search system, place estimation method, and non-transitory computer readable medium storing program |
CN111565300B (zh) * | 2020-05-22 | 2020-12-22 | 深圳市百川安防科技有限公司 | 基于对象的视频文件处理方法、设备及系统 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007174016A (ja) * | 2005-12-20 | 2007-07-05 | Matsushita Electric Ind Co Ltd | 情報配信システム |
WO2013179335A1 (ja) * | 2012-05-30 | 2013-12-05 | 株式会社 日立製作所 | 監視カメラ制御装置及び映像監視システム |
WO2014171258A1 (ja) * | 2013-04-16 | 2014-10-23 | 日本電気株式会社 | 情報処理システム、情報処理方法及びプログラム |
Family Cites Families (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3999561B2 (ja) | 2002-05-07 | 2007-10-31 | 松下電器産業株式会社 | 監視システムと監視カメラ |
JP4929102B2 (ja) | 2007-08-28 | 2012-05-09 | 株式会社東芝 | 映像検索システム及び映像検索装置 |
TWI405457B (zh) * | 2008-12-18 | 2013-08-11 | Ind Tech Res Inst | 應用攝影機換手技術之多目標追蹤系統及其方法,與其智慧節點 |
JP2011170565A (ja) | 2010-02-17 | 2011-09-01 | Toshiba Tec Corp | 顧客動向収集方法、装置及びプログラム |
WO2013108686A1 (ja) | 2012-01-17 | 2013-07-25 | ソニー株式会社 | 情報処理装置および方法、並びにプログラム |
JP2013192154A (ja) | 2012-03-15 | 2013-09-26 | Omron Corp | 監視装置、信頼度算出プログラム、および信頼度算出方法 |
US9589192B2 (en) * | 2012-09-27 | 2017-03-07 | Nec Corporation | Information processing system, information processing method, and program |
AU2013339935A1 (en) * | 2012-10-29 | 2015-05-07 | Nec Corporation | Information processing system, information processing method, and program |
FR2999706B1 (fr) * | 2012-12-14 | 2015-03-20 | Commissariat Energie Atomique | Dispositif de mesure de couple moteur dans un cycle |
WO2014103673A1 (ja) * | 2012-12-28 | 2014-07-03 | 日本電気株式会社 | 情報処理システム、情報処理方法及びプログラム |
US10347100B2 (en) * | 2013-04-26 | 2019-07-09 | Nec Corporation | Information processing system, method and computer readable medium for determining whether moving bodies appearing in first and second videos are the same or not |
WO2015025490A1 (ja) | 2013-08-21 | 2015-02-26 | 日本電気株式会社 | 店舗内顧客行動分析システム、店舗内顧客行動分析方法および店舗内顧客行動分析プログラム |
JP6520783B2 (ja) | 2016-03-22 | 2019-05-29 | 株式会社デンソー | 車両検知装置 |
-
2016
- 2016-09-12 WO PCT/JP2016/004148 patent/WO2017163282A1/ja active Application Filing
- 2016-09-12 EP EP16895322.2A patent/EP3435665A4/en not_active Withdrawn
- 2016-09-12 JP JP2018506503A patent/JP6688975B2/ja active Active
-
2018
- 2018-09-24 US US16/139,527 patent/US10863142B2/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007174016A (ja) * | 2005-12-20 | 2007-07-05 | Matsushita Electric Ind Co Ltd | 情報配信システム |
WO2013179335A1 (ja) * | 2012-05-30 | 2013-12-05 | 株式会社 日立製作所 | 監視カメラ制御装置及び映像監視システム |
WO2014171258A1 (ja) * | 2013-04-16 | 2014-10-23 | 日本電気株式会社 | 情報処理システム、情報処理方法及びプログラム |
Non-Patent Citations (1)
Title |
---|
See also references of EP3435665A4 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11250251B2 (en) * | 2018-09-06 | 2022-02-15 | Nec Corporation | Method for identifying potential associates of at least one target person, and an identification device |
US20220114826A1 (en) * | 2018-09-06 | 2022-04-14 | Nec Corporation | Method for identifying potential associates of at least one target person, and an identification device |
CN110049290A (zh) * | 2019-04-24 | 2019-07-23 | 北京市亚太安设备安装有限责任公司 | 公共建筑物智能监控系统 |
JP2021093052A (ja) * | 2019-12-12 | 2021-06-17 | 清水建設株式会社 | 情報修正装置、情報修正方法およびプログラム |
JP7355632B2 (ja) | 2019-12-12 | 2023-10-03 | 清水建設株式会社 | 情報修正装置、情報修正方法およびプログラム |
JP2022051683A (ja) * | 2020-09-22 | 2022-04-01 | グラスパー テクノロジーズ エーピーエス | 訓練データの生成と再識別に使用するための機械学習モデルの訓練とについての概念 |
JP7186269B2 (ja) | 2020-09-22 | 2022-12-08 | グラスパー テクノロジーズ エーピーエス | 訓練データの生成と同一物判定に使用するための機械学習モデルの訓練とについての概念 |
CN112689132A (zh) * | 2021-03-15 | 2021-04-20 | 成都点泽智能科技有限公司 | 目标对象监控方法和监控设备 |
Also Published As
Publication number | Publication date |
---|---|
US10863142B2 (en) | 2020-12-08 |
JP6688975B2 (ja) | 2020-04-28 |
JPWO2017163282A1 (ja) | 2018-12-13 |
EP3435665A4 (en) | 2019-03-20 |
US20190132556A1 (en) | 2019-05-02 |
EP3435665A1 (en) | 2019-01-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2017163282A1 (ja) | 監視装置及び監視システム | |
US10417773B2 (en) | Method and apparatus for detecting object in moving image and storage medium storing program thereof | |
US9213896B2 (en) | Method for detecting and tracking objects in image sequences of scenes acquired by a stationary camera | |
JP4700477B2 (ja) | 移動体監視システムおよび移動体特徴量算出装置 | |
US10659676B2 (en) | Method and apparatus for tracking a moving subject image based on reliability of the tracking state | |
JP6540330B2 (ja) | 追跡システム、追跡方法および追跡プログラム | |
US9615019B2 (en) | Image capturing apparatus and control method for image capturing apparatus with particle filter for main object detection and selecting focus detection area based on priority | |
JP6280020B2 (ja) | 移動物体追跡装置 | |
US20140198229A1 (en) | Image pickup apparatus, remote control apparatus, and methods of controlling image pickup apparatus and remote control apparatus | |
US10997398B2 (en) | Information processing apparatus, authentication system, method of controlling same, and medium | |
JP2019204518A (ja) | 人物追跡方法、装置、機器及び記憶媒体 | |
CN110651274A (zh) | 可移动平台的控制方法、装置和可移动平台 | |
WO2017163283A1 (ja) | 監視装置及び監視システム | |
US10839552B2 (en) | Image processing apparatus, tracking method, and program | |
JP6833483B2 (ja) | 被写体追尾装置、その制御方法、および制御プログラム、並びに撮像装置 | |
JP6798609B2 (ja) | 映像解析装置、映像解析方法およびプログラム | |
US11074696B2 (en) | Image processing device, image processing method, and recording medium storing program | |
JP2019096062A (ja) | 物体追跡装置、物体追跡方法、および、物体追跡プログラム | |
CN106254861A (zh) | 摄像头的检验方法和装置 | |
JP2010056720A (ja) | ネットワークカメラおよびネットワークカメラシステム | |
JP5539565B2 (ja) | 撮像装置及び被写体追跡方法 | |
JP6548306B2 (ja) | カメラの撮影画像に映る人物を追跡する画像解析装置、プログラム及び方法 | |
US11507768B2 (en) | Information processing apparatus, information processing method, and storage medium | |
JP2005326966A (ja) | 人数カウント装置 | |
CN110706250B (zh) | 一种对象的跟踪方法、装置、系统及存储介质 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 2018506503 Country of ref document: JP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2016895322 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2016895322 Country of ref document: EP Effective date: 20181025 |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 16895322 Country of ref document: EP Kind code of ref document: A1 |