WO2006028106A1 - 異常動作検出装置および異常動作検出方法 - Google Patents
異常動作検出装置および異常動作検出方法 Download PDFInfo
- Publication number
- WO2006028106A1 WO2006028106A1 PCT/JP2005/016380 JP2005016380W WO2006028106A1 WO 2006028106 A1 WO2006028106 A1 WO 2006028106A1 JP 2005016380 W JP2005016380 W JP 2005016380W WO 2006028106 A1 WO2006028106 A1 WO 2006028106A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- feature data
- frame
- abnormal
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING SYSTEMS, e.g. PERSONAL CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0438—Sensor means for detecting
- G08B21/0476—Cameras to detect unsafe condition, e.g. video cameras
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING SYSTEMS, e.g. PERSONAL 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
- G08B13/1961—Movement detection not involving frame subtraction, e.g. motion detection on the basis of luminance changes in the image
Definitions
- the present invention relates to an abnormal operation detection device and an abnormal operation detection method for capturing a moving image and detecting an unusual operation.
- the three-dimensional high-order local autocorrelation feature means that the local autocorrelation feature is calculated at each point in the botacel data (3D data) in which images are arranged in time series. It is a statistical feature obtained by integrating over the entire Botacel data, and can be said to be an operation feature. The recognition result obtained by applying the four actions to discriminant analysis for this feature is very high at almost 100%.
- Non-Patent Document 1 T. Kobayashi ana Action and simultaneous Multiple-Person Identification Using Cubic Higher-Order Local Auto-Correlation "Proceeding of 17th International Conference on Pattern Recognition ⁇ 2004
- An object of the present invention is to solve such problems and to provide an abnormal motion detection device and an abnormal motion detection method for detecting abnormal motion using a three-dimensional higher-order local autocorrelation feature, which is a feature extraction of moving image force and others. It is to provide.
- the abnormal motion detection apparatus of the present invention includes difference data generation means for generating inter-frame difference data from moving image data composed of a plurality of image frame data, and inter-frame difference data power feature data based on higher-order local autocorrelation.
- the feature data extracting means for extracting a plurality of feature data extracted by the feature data extracting means in the past, the subspace based on the principal component vector obtained by the principal component analysis method, and the feature data extracting means
- a distance calculating means for calculating a distance from the feature data, an abnormality determining means for determining an abnormality when the distance is greater than a predetermined value, and outputting a determination result when the abnormality determining means determines an abnormality.
- the main feature is the provision of output means.
- the feature data extraction unit may extract feature data from a plurality of nearest three-dimensional data that is the inter-frame difference data force by a cubic higher-order local autocorrelation. .
- the abnormal operation detection device described above further includes a capturing unit that captures moving image frame data in real time, a frame data storage unit that stores the captured frame data, and feature data extracted by the feature data extraction unit.
- a storage unit that stores data for a certain period of time, and a subspace in which subspace information is updated by obtaining a subspace based on a principal component vector obtained by a principal component analysis method from feature data stored in the storage unit It may be possible to provide an updating means.
- the abnormal motion detection method of the present invention includes a first step of generating moving image data force inter-frame difference data composed of a plurality of image frame data, and a plurality of the three-dimensional data composed of the plurality of inter-frame difference data forces.
- Feature data by 3D higher order local autocorrelation A second step of extracting, a third step of calculating a distance between the subspace based on the principal component vector obtained by the principal component analysis method from a plurality of feature data extracted in the past and the feature data, and the distance is predetermined.
- the main feature is that it has a fourth step for determining an abnormality when it is greater than the value and a fifth step for outputting a determination result when it is determined to be abnormal.
- the first step includes a step of capturing moving image frame data in real time and a step of storing the captured frame data
- the third step includes feature data extraction.
- the feature data extracted by the means is stored for a certain period of time, and the subspace information is updated by obtaining a subspace based on the principal component outline obtained by the principal component analysis method from the stored feature data. Even if you include steps.
- the motion feature does not depend on the position or the like of the target! /, (3D) higher-order local autocorrelation features that are position-invariant values are used, and when there are a plurality of targets, Due to the property that the total feature value is the sum of the individual feature values, normal behavior that is abundantly obtained as normal data is statistically learned as a subspace, and abnormal behavior is derived from it. For example, when there are multiple people on the screen, V, and most of the conventional methods are used to cut out and track individual persons. Can be detected.
- the amount of calculation with a small amount of calculation for feature extraction or abnormality determination is constant regardless of the number of subjects, and there is an effect that real-time processing is possible.
- FIG. 1 is a block diagram showing a configuration of an abnormal operation detection device according to the present invention.
- FIG. 2 is a flowchart showing the contents of abnormal operation detection processing according to the present invention.
- FIG. 3 is a flowchart showing the contents of the S13 cubic high-order local autocorrelation feature extraction process.
- FIG. 4 is an explanatory diagram showing autocorrelation processing coordinates in a three-dimensional pixel space.
- FIG. 5 is an explanatory diagram showing an example of an autocorrelation mask pattern.
- FIG. 6 is an explanatory diagram showing the contents of real-time processing of a moving image according to the present invention.
- FIG. 7 is an explanatory diagram showing the additiveness of CHLAC features and the properties of subspaces.
- FIG. 8 is an explanatory diagram showing an example of the additiveness of CHLAC features and subspaces.
- abnormal operation is that the abnormality itself cannot be defined so that all abnormal events cannot be listed. Therefore, in this specification, abnormal operation is defined as “not normal operation”. Normal operation can be learned from a statistical distribution if it is an operation in which the distribution is concentrated considering the statistical distribution of operation characteristics. An operation that deviates greatly from the distribution is defined as an abnormal operation.
- a general motion such as a walking motion is learned and recognized as a normal motion, but a motion like a suspicious behavior is a walking motion. Since it is not a periodic motion but a small distribution, it is recognized as an abnormal operation.
- the present inventors conducted an experiment with the “walking” operation as a normal operation, the “running” operation, and the “falling” operation as an abnormal operation.
- a partial space of normal motion features is generated in the motion feature space based on the three-dimensional higher-order local autocorrelation features, and the distances of the partial space forces are different. Abnormal operation is detected as a normal value.
- the principal component analysis method is used to generate the normal motion subspace.
- the principal component subspace is composed of principal component vectors with a cumulative contribution rate of 0.99.
- the three-dimensional higher-order local autocorrelation feature has a property that it does not require extraction of an object and is additive in the screen. Due to this additivity, when a normal motion subspace is constructed, the feature vector will fit in the normal motion subspace no matter how many people perform normal motions on the screen, and even one of them will behave abnormally. If it does, it will jump out of the subspace and be detected as an abnormal value. Since there is no need to track and calculate each individual person, the amount of calculation is constant without being proportional to the target number of people and can be calculated at high speed.
- the principal component vector of the CHLAC feature to be learned is obtained, and the force that forms the subspace using the principal component vector.
- the important thing here is that the additive property of the CHLAC feature The compatibility of is very good. Whether or not it belongs to the normal motion subspace (distance is below a predetermined threshold) does not depend on the vector size. That is, only the direction of the vector is an element of whether or not it belongs to the normal operation subspace.
- FIG. 7 is an explanatory diagram showing the additiveness of the CHLAC feature and the properties of the subspace.
- the CHLAC feature data space is 2 dimensions (actually 251 dimensions), and the normal operation subspace is 1 dimension (in the example, for example, if the cumulative contribution ratio is 0.99, 3 ⁇
- the CHLAC feature data of normal operation forms a group for each number of monitoring targets.
- the normal motion subspace S obtained by the principal component analysis exists in the vicinity including the CHLAC feature data of normal operation. Since the CHL AC feature data A of the abnormal operation that deviates from the normal operation subspace S has a large vertical distance d ⁇ , an abnormality is determined by this vertical distance d ⁇ .
- FIG. 8 is an explanatory diagram showing an example of CHLAC feature additivity and subspace.
- Figure 8 (a) shows the CHLAC feature vector for one person's normal motion (walking), and is in the normal motion subspace S (very close).
- Fig. 8 (b) shows the CHLAC characteristic vector for one person's abnormal operation (falling), and is separated from the normal operation subspace S by a vertical distance d ⁇ .
- Fig. 8 (c) shows the CHL AC feature vector when two normal operations (walking) and one abnormal operation (tumbling) coexist, and as in (b). Normal operation subspace Vertical distance from S d ⁇ apart.
- N normal and A is abnormal. Projectors are defined later.
- FIG. 1 is a block diagram showing a configuration of an abnormal operation detection device according to the present invention.
- Video power Mera 10 outputs moving image frame data of the target person or device in real time.
- Video camera 10 can be monochrome or color camera.
- the computer 11 may be, for example, a well-known personal computer (PC) equipped with a video capture circuit for capturing moving images.
- the present invention is realized by creating, installing, and starting up a program to be described later on any known computer 11 such as a personal computer.
- the monitor device 12 is a well-known output device of the computer 11 and is used, for example, to display to the operator that an abnormal operation has been detected.
- the detected anomalies can be reported and displayed on a remote monitoring device via the Internet, etc., by alerting by sound, or by calling a wired or mobile phone and reporting by voice.
- the method of performing etc. is employable.
- the keyboard 13 and the mouse 14 are well-known input devices used for input by the operator.
- moving image data input from the video camera 10 may be processed in real time, or may be stored in a moving image file and sequentially read out for processing.
- FIG. 2 is a flowchart showing the contents of abnormal operation detection processing according to the present invention.
- the process waits until the input of frame data from the video camera 10 is completed.
- S11 In frame data is input (read into a memory).
- the image data at this time is, for example, 256 gray scale data.
- binarization is performed by automatic threshold selection in order to remove color information and noise unrelated to "movement".
- the input moving image data becomes a column of frame data (binary image) having logical values of “moved (1)” and “moved (0)” as pixel values.
- Non-Patent Document 2 Noriyuki Otsu, "Automatic threshold selection method based on discrimination and least squares criterion" IEICE Transactions D, J63-D-4, P349-356, 1980.
- the higher order local autocorrelation function is limited to a local region.
- the displacement direction is limited to a local area of 3 X 3 X 3 pixels centered on the reference point r, that is, 26 neighborhoods of the reference point r.
- the displacement direction in the case of taking a cubic higher-order local autocorrelation feature may not necessarily be separated by adjacent pixels.
- the number of feature amounts that is, the dimension of the feature vector corresponds to the type of mask pattern.
- the number of pixels multiplied by 1 is 1. Therefore, terms with a square or higher are deleted as duplicates of terms with a square that differ only in the multiplier.
- patterns that overlap in the integration operation (translation: scan) in Equation 1 leave one representative pattern and delete others. Since the expression on the right side of Equation 1 always includes the reference point (r) : the center of the local area), select a representative pattern that includes the central point and that fits within the local area of 3 X 3 X 3 pixels. .
- the 3D higher-order local autocorrelation feature vector for one 3D data is 251D.
- the correlation value is a (0th order) ⁇ a X a (l order) ⁇ a X a X a ( Even if the selected pixels are the same, duplicated ones with different multipliers cannot be deleted. Therefore, in the case of multiple values, the number of mask patterns is increased by 2 when the number of selected pixels is 1 and by 26 when the number of selected pixels is 2, compared to the case of 2 values.
- the frame CHLAC data is stored corresponding to the frame.
- the latest CHLAC data obtained in S13 is added to the current CHLAC data, and new CHLAC data is obtained by subtracting the current CHLAC data from the frame CHLAC data corresponding to frames older than a predetermined period. Generate and save.
- FIG. 6 is an explanatory diagram showing the contents of real-time processing of a moving image according to the present invention.
- the moving image data is a sequence of frames. Therefore, a time window with a certain width is set in the time direction, and the set of frames in the window is used as one 3D data. Each time a new frame is input, the time window is moved and the old frame is deleted to obtain finite 3D data. The length of this time window should be set equal to or longer than one period of the motion to be recognized.
- the principal component vectors are obtained by the principal component analysis method from all of the CHLAC data groups stored up to now or the most recent predetermined number, and the normal operation subspace is determined. To do. Since the principal component analysis method itself is well known, an outline will be described. [0039] First, in order to construct a subspace of normal operation, principal component vectors are obtained from the CHLAC feature data group by principal component analysis.
- the M-dimensional CHLAC feature vector X is expressed as follows.
- ⁇ is the average vector of feature vectors X
- the matrix U is obtained from the eigenvalue problem of the following equation using this covariance matrix ⁇ .
- this vertical distance d ⁇ is used as an indicator of whether or not the force is abnormal.
- FIG. 3 is a flowchart showing the contents of the three-dimensional higher-order local autocorrelation feature extraction process in S13.
- S30 251 correlation pattern counters are cleared.
- S31 one unprocessed target pixel (reference point) is selected (scans the target pixel in order in the frame).
- S32 one unprocessed mask pattern is selected.
- FIG. 4 is an explanatory diagram showing the autocorrelation processing coordinates in the three-dimensional pixel space.
- the xy planes of three difference frames of t ⁇ 1 frame, t frame, and t + 1 frame are shown side by side.
- the mask pattern is information indicating the combination of pixels to be correlated, and the pixel data selected by the mask pattern is used to calculate the correlation value, but the powerful pixels not selected by the mask pattern are ignored. .
- the target pixel center pixel
- the number of patterns after eliminating the overlap in a 3 ⁇ 3 ⁇ 3 pixel cube is 251.
- FIG. 5 is an explanatory diagram showing an example of an autocorrelation mask pattern.
- Figure 5 (1) shows the simplest 0th-order mask pattern with only the pixel of interest.
- (2) is an example of a primary mask pattern in which two hatched pixels are selected.
- (3) and (4) are examples of a secondary mask pattern in which three hatched pixels are selected. There are many other patterns.
- Equation 2 is the difference between the coordinates corresponding to the mask pattern and the binarization 3D data pixel value
- Equation 1 corresponds to moving (scanning) the pixel of interest within the frame and adding the correlation values with the counter corresponding to the mask pattern (counting 1).
- S34 it is determined whether or not the correlation value is 1. If the determination result is affirmative, the flow proceeds to S35, but if negative, the flow proceeds to S36. In order to reduce the amount of calculation, the actual calculation first determines whether the pixel value of the reference point is 1 or not after S31 before calculating the correlation value in S33. Even if the correlation is calculated, it is 0, so skip to S37. In S35, the correlation pattern counter corresponding to the mask pattern is incremented by 1. In S36, it is determined whether or not the force has been processed for all patterns. If the determination result is affirmative, the process proceeds to S37. If the determination is negative, the process proceeds to S32.
- S37 it is determined whether or not the processing has been completed for all the pixels. If the determination result is affirmative, the process proceeds to S38. If the determination result is negative, the process proceeds to S31. In S38, a set of pattern counter values is output as 251 dimensional frame CHLAC data.
- normal motion is statistically learned as a subspace, and abnormal motion can be detected as a deviation from it. it can.
- This method can be applied to the case of multiple people, and can be detected if one person performs an abnormal operation on the screen. However, it is an effective and highly practical method that does not require the extraction of the target and the calculation amount is constant regardless of the number of people.
- this method statistically learns normal operations without explicitly defining them, it is in accordance with the monitoring target that does not need to be defined as what normal operation means at the design stage. Natural detection.
- This since no assumptions or knowledge about the monitoring target is required, This is a general-purpose method that can determine normality / abnormality for various monitoring targets.
- online learning can detect abnormal behavior in real time.
- the embodiment for detecting the abnormal operation has been described. However, the following modifications may be considered in the present invention.
- the normal operation partial space is generated in advance by the learning phase, or for example, 1 minute, 1
- a normal operation subspace may be generated and updated at a predetermined period longer than a frame interval such as time or one day, and abnormal operation may be detected using a fixed subspace until the next update. . In this way, the throughput is further reduced.
- Non-patent document 3 Juyang Weng, Yilu Zhang and Wey-bhiuan Hwang, Candid and ovarian ce- Free Incremental Principal Component Analysis ", IEEti, fransactions on Pattern Analysis and Machine Intelligence, Vol.25, No.8, pp.1034— 1040, 2003
- the configuration of the embodiment cannot form the normal operation subspace so correctly, and the detection accuracy of abnormal operations may be reduced. Therefore, after clustering in the normal operation subspace, it is conceivable to measure the distance from the cluster so that it can cope with the multimodal distribution.
- a partial space can be generated for each of a plurality of normal operation patterns
- a plurality of abnormality determinations are performed using each partial space, and a logical product of a plurality of determination results is taken to determine that all patterns are abnormal.
- the determined item may be regarded as abnormal.
- the frame determined to be an abnormal operation is also used for the normal operation partial space generation processing.
- the frame determined to be an abnormal operation may be excluded from the subspace generation processing power. In this way, the detection accuracy is improved when the rate of abnormal operation is high or when the number of image samples is V.
- 3D CHLAC an example of calculating 3D CHLAC was disclosed. However, considering the use, accuracy, and computational complexity, 2D height is calculated from each difference frame that 3D CHLAC does not have. Next, local autocorrelation features may be calculated to generate a subspace of this data force normal motion to detect abnormal motion.
Landscapes
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Human Computer Interaction (AREA)
- Social Psychology (AREA)
- Gerontology & Geriatric Medicine (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| EP05778592A EP1801757A4 (en) | 2004-09-08 | 2005-09-07 | DETECTOR FOR ABNORME ACTION AND DETECTION METHOD FOR ABNORME ACTION |
| US11/662,366 US20080123975A1 (en) | 2004-09-08 | 2005-09-07 | Abnormal Action Detector and Abnormal Action Detecting Method |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2004261179A JP4368767B2 (ja) | 2004-09-08 | 2004-09-08 | 異常動作検出装置および異常動作検出方法 |
| JP2004-261179 | 2004-09-08 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2006028106A1 true WO2006028106A1 (ja) | 2006-03-16 |
Family
ID=36036387
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2005/016380 Ceased WO2006028106A1 (ja) | 2004-09-08 | 2005-09-07 | 異常動作検出装置および異常動作検出方法 |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US20080123975A1 (ja) |
| EP (1) | EP1801757A4 (ja) |
| JP (1) | JP4368767B2 (ja) |
| WO (1) | WO2006028106A1 (ja) |
Families Citing this family (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4701100B2 (ja) * | 2006-02-17 | 2011-06-15 | 株式会社日立製作所 | 異常行動検知装置 |
| JP4215781B2 (ja) * | 2006-06-16 | 2009-01-28 | 独立行政法人産業技術総合研究所 | 異常動作検出装置および異常動作検出方法 |
| JP4603512B2 (ja) * | 2006-06-16 | 2010-12-22 | 独立行政法人産業技術総合研究所 | 異常領域検出装置および異常領域検出方法 |
| JP4429298B2 (ja) * | 2006-08-17 | 2010-03-10 | 独立行政法人産業技術総合研究所 | 対象個数検出装置および対象個数検出方法 |
| JP5121258B2 (ja) | 2007-03-06 | 2013-01-16 | 株式会社東芝 | 不審行動検知システム及び方法 |
| JP4842197B2 (ja) * | 2007-04-17 | 2011-12-21 | 財団法人ソフトピアジャパン | 多重分割画像を用いた異常動作検出装置、異常動作検出方法及び異常動作検出用プログラム |
| JP4769983B2 (ja) * | 2007-05-17 | 2011-09-07 | 独立行政法人産業技術総合研究所 | 異常検出装置および異常検出方法 |
| JP4573857B2 (ja) * | 2007-06-20 | 2010-11-04 | 日本電信電話株式会社 | 逐次更新型非定常検出装置、逐次更新型非定常検出方法、逐次更新型非定常検出プログラムおよびそのプログラムを記録した記録媒体 |
| JP4925120B2 (ja) * | 2007-07-02 | 2012-04-25 | 独立行政法人産業技術総合研究所 | 物体認識装置および物体認識方法 |
| JP4654347B2 (ja) * | 2007-12-06 | 2011-03-16 | 株式会社融合技術研究所 | 異常動作監視装置 |
| JP4953211B2 (ja) * | 2007-12-13 | 2012-06-13 | 独立行政法人産業技術総合研究所 | 特徴抽出装置および特徴抽出方法 |
| WO2010111748A1 (en) * | 2009-04-01 | 2010-10-07 | Curtin University Of Technology | Systems and methods for detecting anomalies from data |
| DE102009021765A1 (de) * | 2009-05-18 | 2010-11-25 | Deutsches Zentrum für Luft- und Raumfahrt e.V. | Verfahren zum automatischen Erkennen einer Situationsänderung |
| JP2011048547A (ja) * | 2009-08-26 | 2011-03-10 | Toshiba Corp | 異常行動検知装置、監視システム及び異常行動検知方法 |
| JP5190968B2 (ja) * | 2009-09-01 | 2013-04-24 | 独立行政法人産業技術総合研究所 | 動画像の圧縮方法及び圧縮装置 |
| JP5131863B2 (ja) * | 2009-10-30 | 2013-01-30 | 独立行政法人産業技術総合研究所 | Hlac特徴量抽出方法、異常検出方法及び装置 |
| CN102317978B (zh) * | 2009-12-22 | 2014-11-12 | 松下电器产业株式会社 | 动作解析装置以及动作解析方法 |
| JP5455787B2 (ja) * | 2010-05-27 | 2014-03-26 | パナソニック株式会社 | 動作解析装置および動作解析方法 |
| JP5675229B2 (ja) | 2010-09-02 | 2015-02-25 | キヤノン株式会社 | 画像処理装置及び画像処理方法 |
| AU2012288992B2 (en) * | 2011-07-28 | 2015-03-12 | Paul Scherrer Institut | Method for image fusion based on principal component analysis |
| US9824296B2 (en) | 2011-11-10 | 2017-11-21 | Canon Kabushiki Kaisha | Event detection apparatus and event detection method |
| CN103106394A (zh) * | 2012-12-24 | 2013-05-15 | 厦门大学深圳研究院 | 一种视频监控中的人体行为识别方法 |
| CN103902966B (zh) * | 2012-12-28 | 2018-01-05 | 北京大学 | 基于序列时空立方体特征的视频交互事件分析方法及装置 |
| CN103310463B (zh) * | 2013-06-18 | 2016-01-20 | 西北工业大学 | 基于概率主成分分析和压缩感知的在线目标跟踪方法 |
| JP6285116B2 (ja) * | 2013-07-02 | 2018-02-28 | Necプラットフォームズ株式会社 | 動作評価装置、動作評価方法及び動作評価プログラム |
| WO2015198767A1 (ja) * | 2014-06-27 | 2015-12-30 | 日本電気株式会社 | 異常検知装置及び異常検知方法 |
| JP6464635B2 (ja) * | 2014-09-26 | 2019-02-06 | 株式会社リコー | 画像処理装置、画像処理装置の制御方法、及びプログラム |
| WO2016079833A1 (ja) * | 2014-11-19 | 2016-05-26 | 富士通株式会社 | 異常検出装置、異常検出方法および異常検出プログラム |
| WO2016088758A1 (ja) * | 2014-12-01 | 2016-06-09 | 国立研究開発法人産業技術総合研究所 | 超音波検査システム及び超音波検査方法 |
| CN106157326B (zh) * | 2015-04-07 | 2020-10-27 | 中国科学院深圳先进技术研究院 | 群体异常行为检测方法及系统 |
| CN107949865B (zh) * | 2015-09-02 | 2021-10-29 | 富士通株式会社 | 异常检测装置、异常检测方法 |
| JP6708385B2 (ja) | 2015-09-25 | 2020-06-10 | キヤノン株式会社 | 識別器作成装置、識別器作成方法、およびプログラム |
| JP6336952B2 (ja) * | 2015-09-30 | 2018-06-06 | セコム株式会社 | 群衆解析装置 |
| SG10201510337RA (en) | 2015-12-16 | 2017-07-28 | Vi Dimensions Pte Ltd | Video analysis methods and apparatus |
| CN105513095B (zh) * | 2015-12-30 | 2019-04-09 | 山东大学 | 一种行为视频无监督时序分割方法 |
| JP6702340B2 (ja) * | 2016-01-28 | 2020-06-03 | 株式会社リコー | 画像処理装置、撮像装置、移動体機器制御システム、画像処理方法、及びプログラム |
| JP6884517B2 (ja) | 2016-06-15 | 2021-06-09 | キヤノン株式会社 | 情報処理装置、情報処理方法及びプログラム |
| JP7035395B2 (ja) * | 2017-09-13 | 2022-03-15 | 沖電気工業株式会社 | 異常検知システム、情報処理装置、及び、異常検知方法 |
| CN111149129A (zh) | 2017-09-21 | 2020-05-12 | 株式会社安德科技术 | 异常检测装置及异常检测方法 |
| CN110276398A (zh) * | 2019-06-21 | 2019-09-24 | 北京滴普科技有限公司 | 一种视频异常行为自动判断方法 |
| US11526958B2 (en) | 2019-06-26 | 2022-12-13 | Halliburton Energy Services, Inc. | Real-time analysis of bulk material activity |
| CN112822434A (zh) * | 2019-11-15 | 2021-05-18 | 西安科芮智盈信息技术有限公司 | 防窃照处理方法、设备及系统 |
| KR102770773B1 (ko) * | 2022-07-25 | 2025-02-24 | 주식회사 아임토리 | 인공지능 기반의 로봇 팔 이미지 데이터 처리 방법 |
| CN116884187B (zh) * | 2023-06-25 | 2026-02-06 | 吉林金域医学检验所有限公司 | 实验室监测方法、装置、电子设备及存储介质 |
Family Cites Families (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0363828B1 (en) * | 1988-10-11 | 1999-01-07 | Kabushiki Kaisha Ouyo Keisoku Kenkyusho | Method and apparatus for adaptive learning type general purpose image measurement and recognition |
| US6545115B2 (en) * | 1996-12-11 | 2003-04-08 | Rhodia Chimie | Process for preparing a stable silicone oil containing SiH groups and hydrosilylable functions |
| US6466685B1 (en) * | 1998-07-14 | 2002-10-15 | Kabushiki Kaisha Toshiba | Pattern recognition apparatus and method |
| JP4226730B2 (ja) * | 1999-01-28 | 2009-02-18 | 株式会社東芝 | 物体領域情報生成方法及び物体領域情報生成装置並びに映像情報処理方法及び情報処理装置 |
| EP1297691A2 (en) * | 2000-03-07 | 2003-04-02 | Sarnoff Corporation | Camera pose estimation |
| US7522186B2 (en) * | 2000-03-07 | 2009-04-21 | L-3 Communications Corporation | Method and apparatus for providing immersive surveillance |
| US7110569B2 (en) * | 2001-09-27 | 2006-09-19 | Koninklijke Philips Electronics N.V. | Video based detection of fall-down and other events |
| US7016884B2 (en) * | 2002-06-27 | 2006-03-21 | Microsoft Corporation | Probability estimate for K-nearest neighbor |
| CN100477745C (zh) * | 2002-08-09 | 2009-04-08 | 夏普株式会社 | 图像合成装置和图象合成方法 |
| JP3954484B2 (ja) * | 2002-12-12 | 2007-08-08 | 株式会社東芝 | 画像処理装置およびプログラム |
| JP4090388B2 (ja) * | 2003-05-27 | 2008-05-28 | 三洋電機株式会社 | 電波受信装置 |
| US7616782B2 (en) * | 2004-05-07 | 2009-11-10 | Intelliview Technologies Inc. | Mesh based frame processing and applications |
| US20060018516A1 (en) * | 2004-07-22 | 2006-01-26 | Masoud Osama T | Monitoring activity using video information |
| JP3970877B2 (ja) * | 2004-12-02 | 2007-09-05 | 独立行政法人産業技術総合研究所 | 追跡装置および追跡方法 |
| US7739284B2 (en) * | 2005-04-20 | 2010-06-15 | International Business Machines Corporation | Method and apparatus for processing data streams |
| US7376246B2 (en) * | 2005-06-27 | 2008-05-20 | Mitsubishi Electric Research Laboratories, Inc. | Subspace projection based non-rigid object tracking with particle filters |
| US7760911B2 (en) * | 2005-09-15 | 2010-07-20 | Sarnoff Corporation | Method and system for segment-based optical flow estimation |
| JP4603512B2 (ja) * | 2006-06-16 | 2010-12-22 | 独立行政法人産業技術総合研究所 | 異常領域検出装置および異常領域検出方法 |
| JP4215781B2 (ja) * | 2006-06-16 | 2009-01-28 | 独立行政法人産業技術総合研究所 | 異常動作検出装置および異常動作検出方法 |
| JP4429298B2 (ja) * | 2006-08-17 | 2010-03-10 | 独立行政法人産業技術総合研究所 | 対象個数検出装置および対象個数検出方法 |
-
2004
- 2004-09-08 JP JP2004261179A patent/JP4368767B2/ja not_active Expired - Lifetime
-
2005
- 2005-09-07 EP EP05778592A patent/EP1801757A4/en not_active Withdrawn
- 2005-09-07 WO PCT/JP2005/016380 patent/WO2006028106A1/ja not_active Ceased
- 2005-09-07 US US11/662,366 patent/US20080123975A1/en not_active Abandoned
Non-Patent Citations (3)
| Title |
|---|
| KOBAYASHI T.ET AL: "Action and Simultaneous Multiple-Person Identification Using Cubic Higher-order Local Auto-Correlation", PROCEEDINGS OF THE 17TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR'04), vol. 4, 2004, pages 741 - 744, XP010724027 * |
| See also references of EP1801757A4 * |
| YOSUKE YAMADA ET AL: "Koji Kyokusho Jiho Sokan Tokucho to Hanbetsu Bunseki ni yoru Kao Gazo no Hyojo Ninshiki. (The Expression Recognition......)", ITE TECHNICAL REPORT, 2003 NEN 2 GATU KAISAIBUN, THE INSTITUTE OF IMAGE INFORMATION AND TELEVISION ENGINEERS, vol. 27, no. 8, 3 February 2003 (2003-02-03), pages 145 - 148, XP002998978 * |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2006079272A (ja) | 2006-03-23 |
| JP4368767B2 (ja) | 2009-11-18 |
| US20080123975A1 (en) | 2008-05-29 |
| EP1801757A1 (en) | 2007-06-27 |
| EP1801757A4 (en) | 2012-02-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2006028106A1 (ja) | 異常動作検出装置および異常動作検出方法 | |
| JP4215781B2 (ja) | 異常動作検出装置および異常動作検出方法 | |
| CN109076198B (zh) | 基于视频的对象跟踪遮挡检测系统、方法和设备 | |
| US7916944B2 (en) | System and method for feature level foreground segmentation | |
| JP4842197B2 (ja) | 多重分割画像を用いた異常動作検出装置、異常動作検出方法及び異常動作検出用プログラム | |
| JP3970877B2 (ja) | 追跡装置および追跡方法 | |
| Avgerinakis et al. | Recognition of activities of daily living for smart home environments | |
| KR101750094B1 (ko) | 영상의 실시간 모니터링을 통한 비정상 집단행동 분류 방법 | |
| JP4061377B2 (ja) | 3次元データからの特徴抽出装置 | |
| US20100195899A1 (en) | Detection of people in real world videos and images | |
| EP2352128A1 (en) | Mobile body detection method and mobile body detection apparatus | |
| Poonsri et al. | Improvement of fall detection using consecutive-frame voting | |
| Moore | A real-world system for human motion detection and tracking | |
| Wang et al. | Experiential sampling for video surveillance | |
| Siva et al. | Real-time, embedded scene invariant crowd counting using scale-normalized histogram of moving gradients (HoMG) | |
| JP5186656B2 (ja) | 動作評価装置および動作評価方法 | |
| Villamizar et al. | Watchnet: Efficient and depth-based network for people detection in video surveillance systems | |
| Xie et al. | Recognizing violent activity without decoding video streams | |
| Saxena et al. | Video inpainting detection and localization using inconsistencies in optical flow | |
| Biswas et al. | Short local trajectory based moving anomaly detection | |
| JP2014182585A (ja) | 画像解析装置、画像解析方法および画像解析プログラム | |
| CN102254329A (zh) | 基于运动向量归类分析的异常行为检测方法 | |
| JP2015049702A (ja) | 対象認識装置、対象認識方法、およびプログラム | |
| Mantini et al. | A signal detection theory approach for camera tamper detection | |
| Iqbal et al. | On video based human abnormal activity detection with histogram of oriented gradients |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS KE KG KM KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NG NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SM SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
| AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU LV MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 2005778592 Country of ref document: EP |
|
| WWP | Wipo information: published in national office |
Ref document number: 2005778592 Country of ref document: EP |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 11662366 Country of ref document: US |
|
| WWP | Wipo information: published in national office |
Ref document number: 11662366 Country of ref document: US |

