JP6526953B2 - Object identification method - Google Patents

Description

  The present invention relates to an apparatus for detecting an object and identifying its type, and a method using the apparatus.

  In recent years, surveillance cameras for the purpose of crime prevention and crime investigation support and in-vehicle cameras for the purpose of driving assistance have been developed and disseminated. Conventionally, in surveillance cameras, a scene of a crime or a criminal is found by checking a huge image by visual inspection of a person. In order to reduce the effort of visual confirmation of human beings, in addition to showing captured and recorded images to police officers, security guards and users, there is also a function to automatically detect moving objects by image processing. .

  As a technique for automatically detecting a person from an image, a method described in Patent Document 1 has been proposed. This detects the area of the moving object from the difference between the past image and the current image, and tracks the moving object, thereby removing plants swaying in the wind, etc. and detecting only people who move continuously. It is a thing.

JP 2002-157599 A

  However, according to the method described in Patent Document 1, it is possible to prevent false detection in which an object that does not move as large as a plant swaying in the wind is detected as a person can be prevented. There is a problem that the type of the continuously moving object can not be identified.

  As one mode for solving the above problem, an object identification apparatus according to the present invention mainly determines an object detection unit that detects an object moving in a moving image, and a flow line of the object detected by the object detection unit. A motion line calculation unit, and an object identification unit that identifies the type of the object or analyzes the attribute of the object by analyzing the shape of the movement line obtained by the motion line calculation unit.

  According to the above aspect, since it is possible to provide a device capable of identifying the type and attribute of the moving object and a method using the device in addition to detecting the moving object, it is possible to support crime prevention and criminal investigation. By identifying the type of the object with the monitoring camera and presenting only the image in which the specific object is shown to the user, it is possible to further reduce the time required for visual confirmation of the person.

It is a figure which shows the example of the basic composition of the object identification device which concerns on this Embodiment. It is a figure which shows the example of the flow line calculation result which concerns on this Embodiment. It is a figure which shows the other example of the flow line calculation result which concerns on this Embodiment. It is a figure which shows the other than the flow line calculation result which concerns on this Embodiment. It is a figure which shows the other example of the flow line calculation result which concerns on this Embodiment. It is a figure showing the example of the basic composition of the image processing device to which the object identification device concerning this embodiment is applied. It is a figure which shows the example of a structure of the image processing system to which the object identification apparatus which concerns on this Embodiment is applied. It is a figure which shows the example of a structure of the image storage system to which the object identification apparatus which concerns on this Embodiment is applied. It is a figure which shows the example of a structure of the image search system to which the object identification apparatus which concerns on this Embodiment is applied. It is a figure which shows the other example of a structure of the image processing apparatus to which the object identification apparatus which concerns on this Embodiment is applied. It is a figure which shows the example of the processing flow of the whole object identification apparatus which concerns on this Embodiment. It is a figure which shows the example of the processing flow of the flow line calculation part which concerns on this Embodiment. It is a figure explaining the principle of three-dimensional measurement of the stereo camera concerning this embodiment. It is a figure which shows the example of the processing flow of the object identification part which concerns on this Embodiment. It is a figure which shows the labeling of the object which concerns on this Embodiment, and the example of a circumscribed rectangle. It is a figure explaining the example of person decision using the feature-value which concerns on this Embodiment.

Hereinafter, the present embodiment will be described with reference to the drawings.
First Embodiment
FIG. 1 is a block diagram showing an example of the basic configuration of an object identification apparatus according to the first embodiment. Details will be described later.

  In the first embodiment, an object identification apparatus that processes an image captured from a camera that is an imaging apparatus and detects an object shown in the image and identifies the type of the object, and image processing to which the apparatus is applied The apparatus will be described.

  First, an image processing apparatus according to the present embodiment will be described with reference to FIG.

  In the image processing apparatus 41 as shown in FIG. 6, the object identification apparatus 1 according to the present embodiment is applied to, for example, the CPU 6 or a unit corresponding thereto, and the camera 4 (imaging apparatus) included in the image processing apparatus 41 A function of capturing an image of the captured surrounding environment and processing the captured image to detect an object can be realized.

  The image processing apparatus 41 includes a camera 4 as an imaging device as shown in FIG. 6, a CPU 6 as processing means for processing an image picked up by the camera 4, a memory 9 as recording means for the CPU 6, The ROM 7 is a data recording means, and an image input means 5 for controlling the camera 4 and taking in an image, which are connected by a bus 11.

  The camera 4 of the image processing apparatus 41 mounts, for example, a CCD or a CMOS imaging device. The image input means 5 stores the image picked up by the camera 4 in the memory 9 and the CPU 6 processes the image based on the program and data stored in the ROM 7. The processing result is displayed on the monitor 16 connected to the I / F 14, and when the object is detected, a warning sound is emitted from the speaker 17 or the like to notify the user that the object has been detected. The input device 15 is also for changing settings of the image processing apparatus 41 and the like. However, this configuration is not an essential requirement, and in addition to the CPU 6, a configuration in which an image processing LSI is mounted and a part of processing is processed by a dedicated LSI, and not only one memory but also a RAM for the image processing LSI Alternatively, a configuration using a plurality of memories may be used. Alternatively, a storage device such as a hard disk not shown may be used instead of the ROM 7.

  Further, as shown in FIG. 7, the camera unit 42 and the processing unit 43 may be provided. In this case, the camera unit 42 communicates with the processing unit through the I / F 13 via the data transmission means 3 the information such as the camera shutter speed, gain and focal length necessary for the image and processing from the I / F 12. Furthermore, as shown in FIG. 8, an image recording system 45 for storing an image for a long time can be configured by further providing the recording device 31. Then, as shown in FIG. 9, the recording device 31 is connected to a computer or other I / F 18 such as a USB, and the recorded image is processed to identify the type and attribute of the object, and the object of the specific type and attribute is automatically identified. It may be a configuration of the image search system 44 that searches for an image in which the image is displayed.

  Next, the configuration of the object identification device shown in FIG. 1 will be described as an example. The object identification device 1 mainly executes the program executed by the CPU 6 of the image processing device 41 described above with reference to FIG. 6, and processes an moving image acquired from the camera 4 to detect an object moving. It comprises a movement line calculation unit 21 which is a movement trajectory of the detected object, and an object identification unit 22 which analyzes the shape of the movement line and identifies the type or attribute of the object.

  The object detection unit 20 has a function of detecting an object moving by a known background difference, an inter-frame difference, or the like.

  The movement line calculation unit 21 has a function of calculating a series of the center of gravity of the object area as a movement line, and obtaining a movement line by associating feature points in the object area with images having different imaging timings by tracking processing.

  The object identification unit 22 includes the frequency and amplitude of the flow line, the variation of individual line segments constituting the flow line, the angle formed by the average movement direction and the individual line segments constituting the flow line, and the line segments constituting the flow line. It has a function of identifying the type of an object from a series or the like.

  The entire processing of the object identification device according to the present embodiment will be described along the flowchart shown in FIG. First, creation of a past image, clearing of flow line information, etc. are initialized (step 101). In step 102, it is judged whether or not it is the initialization timing when the lighting conditions are largely changed, the installation position and the angle of the camera are changed during the continuous operation, and the initialization timing If it is (step 103), the initialization process is executed other than immediately after system startup (step 111). Then, in step 112, the moving object is detected by the difference between the past image (for example, time t-1) and the current (time t) image. In step 113, a corner having a change in luminance in the vertical and horizontal directions is detected as a feature point from the image at time t-1 using a known Harris detector, Susan detector or the like. Alternatively, the head or the like may be detected, and the center or the like of the detected head region may be used as the feature point. Of course, other than the head, a moving body, an arm, a leg, etc. may be used. At this time, when the tracking is successful at time t-1, the point obtained by the tracking is used as it is, and when the tracking fails, the feature point which has been tracked up to now is removed and the newly detected feature point is Add as a tracking point.

  Next, the coordinates of the feature point on the image at time t are determined from the feature points on the image at time t-1. By sequentially repeating this process, as shown in FIG. 2, it is possible to obtain a flow line that a person walks in, and a flow line in which a two-wheeler has moved (step 114). Then, the calculated length of the flow line is compared with the threshold value, and it is determined whether the length of the flow line is sufficient (step 120). If it is determined that the length is sufficient, step 115 is performed. Transition.

  In step 115, the shape of the flow line is analyzed, and the amplitude of the flow line, the frequency, the average direction of the flow line and the variation of each line segment constituting the flow line, the average direction of the flow line and each line segment constituting the flow line At least one of the angles is determined as a feature value.

  In step 116, an object is identified based on the information obtained in step 115. For example, if the amplitude is equal to or higher than the threshold value, it is a person, and if less than the threshold value, two wheels are used. Alternatively, identification may be made by known learning of neural network, SVM (Support Vector Machine), machine learning such as random forest, discriminant analysis or the like using the learning result 23 learned in advance.

  Step 117 verifies whether or not the object identification result is correct in consideration of the past identification result, and determines the identification result. Then, in step 118, a flag corresponding to the object type is set. For example, in the case of a two-wheeled vehicle, the flag is set to 1, and in the case of a person, the flag is set to 2. Finally, in step 119, the degree of reliability is determined based on how many frames of flow lines have been calculated. For example, when a flow line of 100 frames or more is required, the reliability is 100, and when only a flow line of one frame is obtained, the reliability is 1. Of course, the configuration may be such that the reliability is calculated using an exponential function, logarithm, polynomial, or the like, or the configuration may be such that the flow line length is converted to the reliability using a lookup table.

The process flow of the flow line calculation of step 114 will be described in detail with reference to FIG. The flow line calculation unit 21 of the object identification device 1 shown in FIG. 1 executes this processing flow. First, it is determined whether or not it is an initialization timing such as when the object is out of the field of view of the camera 4 (step 120). If it is the initialization timing (121), the past tracking information is cleared at step 126, and if there is an object whose flow line is to be determined at step 127, a new feature point is registered. In step 122, coordinates of the feature point at time t-1 are determined by a known sum of squared difference (SSD), sum of abusolution difference (SAD), normalized cross-correlation (NCC), gradient method (Lucas-Kanade method) or the like. The correspondence between the coordinates of the feature point at time t and the point t is determined. Here, tracking of feature points (optical flow) using the gradient method will be described. The luminance at a time t of the pixel p = (x, y) in the image is I (x, y, t). It is assumed that this pixel is moved to coordinates (x + Δx, y + Δy) at time t + Δt. Assuming that the values of the pixels of the image do not change after movement, equation (1) holds.
Equation (2) is obtained by Taylor expansion of the right side and ignoring higher order terms.
If both parts are divided by Δt and Δt → 0,
Is obtained.
However,
It is.

Assuming that the displacement amounts u of the respective points in the local region are equal, it is sufficient to find u in which the relational expression of equation (3) is most applicable in the local region, and for each pixel of the local region R shown in equation (5) It is sufficient to find u that minimizes the squared error E.

  In step 123, if the feature point correspondence fails, the process proceeds to step 126, and if successful, the process proceeds to step 124, and the information on the flow line is updated. The information on the flow line is a series of coordinates of feature points from time t-N to time t, and the start frame of the flow line is information such as N frames before, assuming that the flow line is obtained from time tN to time t. The flow line can be obtained by repeating this one by one.

  Here, an example of the process (step 115) of analyzing the shape of the flow line in FIG. 11 will be described using FIGS. 4 and 5. Since the movement line is the movement amount of the feature point between the frames, as shown in FIG. 4, the movement amount at time t and time t-1, the movement amount at time t-1 and time t-2, ..., time It consists of a plurality of line segments of movement amounts of t- (N-1) and time t-N. That is, it is a series of coordinates of N + 1 points. The frequency can be determined by converting the series of coordinates into frequency components by Fourier transform or the like. Alternatively, as shown in FIG. 5, the flow line may be averaged, the distance between the average straight line and the coordinates of the flow line may be determined, and the distance may be used as the amplitude. In addition, the average movement direction of the movement line and the angle θ formed by the line segments of each movement line, the movement direction vector of the movement line average, the size of the movement direction vector of the movement line average, and the movement direction of the movement line average A histogram or the like of the angle θ formed by the line segments of each flow line may be obtained.

  An example of the process (step 116) of identifying the type of the object in FIG. 11 will be described with reference to FIG. Although a random forest will be described by way of example, it may be replaced by a known classifier, analysis method or the like. First, at least one of the frequency, the amplitude, the angle θ formed by the straight line and the line segment of the flow line, the line segment of the flow line, etc. determined in step 115 is used as a feature quantity vector and is input as a classifier. . For example, the feature quantity vector x = (the angle between the moving direction of the average of the frequency amplitude motion line and each line segment of the motion line). The feature quantity vector x is input to L binary decision trees. Each binary decision tree node classifies the object type of the object from the value of each feature amount. By repeating this to the terminal node, it is determined for each binary decision tree which object type the object belongs to. Since this process is performed on L binary decision trees, a histogram y of the determination results obtained by collecting the determination results is created, and the object type with the largest voting value is determined as the identification result. Object type 3 can be identified in the example of FIG. The random forest is also an example shown in FIG. 14 and is not an example of a binary decision tree, but may be a decision tree classified into three or more at each node. Further, the type of the object is, for example, a bicycle, a motorcycle, a pedestrian, a bicycle upper body, a motorcycle upper body, or a pedestrian upper body. There is no restriction on the total number of types of objects to be identified, and an object walking on a foot and an object moving on a wheel, such as a four-legged animal or a car, may be added as an identification target. Also, it is possible to identify an object on which a person is a power source and an object on which a motor is mounted, such as a bicycle and a motorcycle.

  The length of the flow line used for analysis can be determined based on the walking cycle. For example, if it is while advancing one step, it is 0.5 period, and if it is advancing two steps, it is one period. In the case where the moving image has a frame rate of 30 fps (frames per second), it means that it is sufficient to use flow lines of 15 frames or more and 30 frames or more, respectively. In the case of 2 steps in 2 seconds, it is double. Of course, it is possible to execute the identification process even below this. This cycle may be not only walking but also movement of a foot over a bicycle or the like.

  Although the method described above has been described with respect to a camera using one camera, it is apparent that the present embodiment can be applied even when there are two or more cameras. Further, although the case of obtaining the flow line of one point has been described, as shown in FIG. 2B, a plurality of feature points are detected from the object area to obtain a plurality of flow lines, or a head, an arm, a body, a foot , Etc. may be detected to determine the flow line of each part. Finding a plurality of flow lines not only makes the calculation error of the flow lines robust, but also comparing the only part of the object, even if the same movement is performed, using a plurality of flow lines makes the object Can identify the type of For example, assuming that a series of gravity centers of objects is a flow line, even if the flow lines of a two-wheeled vehicle and a human are similar, different flow lines in areas such as feet and hands if a plurality of flow lines are determined in the object area Since it is possible to correctly identify the type of object.

  The two-wheeled vehicle is similar in shape because a person straddles. In particular, the difference is small when moving in a direction approaching or away from the camera. By using the first embodiment, there is an advantage that an object whose shape is similar can be correctly identified.

  Further, as shown in FIG. 3 (a), the contours of the two-wheeled vehicle are similar to the case where only the upper body of a person appears at the lower end of the image, and the conventional HOG (Histogram of Oriented Gradient) feature, Haar In the method using the shape of the object such as -like feature amount, erroneous identification occurs. This problem can also be solved in the first embodiment.

  Furthermore, as shown in FIG. 3B, the upper body of the walking person and the upper body of the person riding a two-wheel vehicle are similar in shape because the two-wheeled vehicle is outside the field of view of the camera. Also in this case, the shape of the flow line is different between a person walking with a foot and a person moving with two wheels, so it can be correctly identified.

  Instead of one flow line for one object, a plurality of flow lines may be obtained for one object as shown in FIG. 2 (b).

  The object to be identified may be, other than a person and a two-wheeled vehicle, a car having a similar aspect ratio, a four-legged animal, a person pushing (pulling) a car and a truck, or the like.

  The object detection process of step 112 has been described by way of example of a method of detecting an object by taking a difference between a past image and a current image, such as a background difference or an inter-frame difference. If the camera is moving, these techniques can not detect the object. In that case, an object may be detected by a classifier such as SVM using a known HOG feature amount and Haar-like feature amount. Alternatively, a stationary object and a moving object can be separated from an optical flow representing the amount of movement of pixels between frames.

  Instead of the optical flow, a moving line used in the present invention may be used to separate a stationary object and a moving object. In that case, the object detection unit 20 calculates flow lines, detects similar flow lines by clustering processing as one object, and separates a stationary object from a moving object. The movement line calculation unit 21 may extract only the movement line included in the area of the moving object detected by the object detection unit 20.

  Although the case of one object is described as an example, the present invention can be applied to the case where there are a plurality of objects. When a plurality of objects are present, as shown in FIG. 15, processing such as assigning different labels to the respective object regions or finding a circumscribed rectangle of each object is performed by a known labeling algorithm. When the coordinates of the feature point at time t are included in the area of label 1 at time t, the flow line can be determined as the flow line of the object to which label 1 is applied. Alternatively, when the coordinates of the feature point at time t are included in the area of the circumscribed rectangle 1, the flow line is determined to be the flow line of the object of the circumscribed rectangle 1.

  Furthermore, based on the result of the object identification unit, an alarm can be issued to the user. For example, when a two-wheeled vehicle is detected, sound is emitted from the speaker 17 and an image showing the detected two-wheeled vehicle is displayed on the monitor 16. Of course, in the case of a person, not a two-wheeled vehicle, an alarm may be issued.

The image search system 44 shown in FIG. 9 searches for an image in which an object of a certain type is shown based on the result of the object identification unit, and gives priority to images with high reliability based on the reliability obtained in step 119. Can be presented to the user. The type of the object can also be designated by the input device 15.
Second Embodiment
The second embodiment will be described in detail with reference to the drawings.

  The second embodiment processes an image captured from a stereo camera which is an imaging device, detects an object shown in the image, and identifies an object type, and an image to which the device is applied. The processing apparatus will be described.

  The object identification apparatus according to the present embodiment is applied to, for example, the CPU 6 in the image processing apparatus 41 as shown in FIG. 10 or a unit corresponding thereto, and a plurality of cameras 4a and 4b included in the image processing apparatus 41. An image of a surrounding environment is processed, and a captured image is processed to realize a function of detecting an object. At this time, three or more cameras may be provided.

  FIG. 13 is a diagram for explaining the principle of three-dimensional measurement of the stereo camera according to the present embodiment. As shown in the figure, when capturing a scene with an overlapping field of view by two cameras with different viewpoints separated by a distance b, the measurement point separated by a distance Z deviates by the parallax δ on the image acquired from the two images Projected on the That is, by obtaining the parallax δ, it is possible to calculate three-dimensional coordinates of points on the image according to the principle of triangulation. At this time, three-dimensional measurement can be performed with high accuracy by correcting the deviation of the positional relationship between the cameras, the deviation of the focal distance of the lenses, the difference between the cameras of the lens distortion, and the like by known camera calibration.

By using a stereo camera that can obtain three-dimensional coordinates, it is possible to obtain three-dimensional flow lines from flow lines on a two-dimensional image. Thereby, for example, the shape of the flow line on the two-dimensional image is different between the person walking sideways and the person walking in the front-rear direction of the camera, but when converted to a three-dimensional space, the shape is different only by the direction. Are similar. The type of the object can be identified with high accuracy by obtaining the feature quantity such as the above-mentioned frequency from the three-dimensional flow line. Alternatively, the flow line of an object moving in the optical axis direction of the camera, the flow line of an object moving in an oblique direction, and the flow line of an object moving in a lateral direction It may be converted by rotation or parallel movement, and may be mapped to a two-dimensional coordinate to obtain a feature quantity such as the amplitude of the flow line.
Third Embodiment
The third embodiment will be described in detail with reference to the drawings.

FIG. 16 is a diagram for explaining an example of person determination using the feature amount according to the present embodiment.
An object identification apparatus according to the present invention is applied to, for example, a CPU 6 or a unit corresponding thereto in an image processing apparatus 41 as shown in FIG. 10, and a plurality of cameras 4a and cameras which are imaging apparatuses included in the image processing apparatus 41. This is an object identification method of processing an image captured from 4b to detect an object shown in the image and identifying the type of the object. At this time, the cameras do not have to face in the same direction.

  When person A and person B pass through the imaging area of camera 4a and then person C and person D enter the imaging area of camera 4b, it is determined whether person A corresponds to person C or person B This makes it possible to track a wide range of people. At this time, the feature amount A obtained by analyzing the shape of the flow line of the person A is identical to the feature amount C obtained by analyzing the shape of the flow line of the person C and the person D and the feature amount D. Identify whether or not a person.

  In the process of step 116, the feature amount A of the flow line calculated from the person A in the camera 4a, the feature amount C of the flow line calculated from the person C in the camera 4b, and the feature amount D of the flow line calculated from the person D It is determined whether the person A and the person C are the same person or whether the person A and the person D are the same person by comparing. For example, if the difference between the amplitude A or frequency A of the flow line of the person A and the amplitude C or frequency C of the flow line of the person C is within the range of the threshold, the same person is used. At this time, in addition to the flow line information, appearance features such as color, gradation, contour, texture, etc. may be used in combination. Or, a histogram A of the angle θ formed by the average moving direction of the movement line of the person A and the line segments of the movement line, and a histogram C of the angle θ formed by the average movement direction of the movement line of the person C and the line segments A known histogram distance (for example, Bhattacharyya Distance) may be obtained, and when the distance is equal to or less than a threshold value, it may be determined as the same person. Alternatively, the histogram D of the angle θ of the movement direction of the human D movement line and each line segment of the movement line is determined, the histogram distance of the histogram A and the histogram C, and the histogram distance of the histogram A and the histogram D are determined. It can be determined that the same person. Of course, if the feature amount of each person can be measured in advance, identification may be performed using a classifier such as SVM or random forest.

Although the case of two cameras, the camera 4a and the camera 4b, has been described, the present method can be applied similarly even if there are three or more cameras. In addition, even if there is only one camera, it is possible to identify whether a person who has passed in the past and a person who has currently passed are the same person.
Fourth Embodiment
The fourth embodiment will be described in detail with reference to the drawings.

  In the fourth embodiment, an object identification apparatus will be described which processes an image captured from a plurality of cameras 4 which are imaging devices, detects an object shown in the image, and identifies an attribute of the object.

  The attributes of an object are, for example, a person holding an umbrella and a person not holding an umbrella. People with age groups such as men and women, children and adults, people with handcuffs and those who do not have handcuffs, types of luggage such as carrying bags or carrying carts, people with high heeled shoes It is also possible to identify a person wearing low-heeled shoes, a person wearing a skirt and a person wearing shorts as attributes. Even in the case of the same object type of human being, the shape of the flow line changes depending on the gender, clothes, and luggage carried, so that the attributes of these objects can be identified by the present invention.

  In the process of identifying the attribute of the object, in the process of step 116, instead of the type of the object, the attribute of the object is learned in advance as an identification class, and the feature amount obtained in step 115 from the input moving image Identifies an attribute The processing is equivalent although there is a difference that the object type 1 to the object type 5 in FIG. 14 become the object attribute 1 to the object attribute 5. Of course, the attribute of the object does not have to be five, and may be more or less.

Although only the feature quantities obtained from the flow lines in step 115 have been described, the gray level and color of the pixels of the image, the change in luminance and color in the object area, and the shape of the moving object area extracted by the background difference and the interframe difference You may use together etc. as a feature-value.

DESCRIPTION OF SYMBOLS 1 ... object identification method, 3 ... data transmission means, 4 ... camera, 5 ... image input means, 6 ... CPU, 7 ... ROM, 9 ... memory, 11 ... · · · · · · Bus, 12 · · · interface (I / F), 13 · · · interface (I / F), 14 · · · interface (I / F), 15 · · · input device, 16 · · · monitor, 17: speaker 18: interface (I / F) 20: object detection unit 21: movement line calculation unit 22: object identification unit 23: learning result 31 ... recording device, 41 ... image processing device, 42 ... camera unit, 43 ... processing unit, 44 ... image search system, 45 ... image recording system,

Claims (11)

An object detection unit that detects an object moving from a moving image;
A movement line calculation unit for calculating a movement line of the object detected by the object detection unit;
An object identification unit for identifying an object being walked by a leg and a two-wheeled vehicle based on the shape of the flow line obtained by the flow line calculation unit ;
An object identification apparatus characterized in that the reliability of the identification result of the type or attribute of an object is calculated based on the length of the flow line obtained by the flow line calculation unit . In the object identification device of claim 1,
An object identification apparatus characterized by identifying an object moving with a wheel and an object walking with the leg. In the object identification device of claim 1,
An object identification apparatus characterized by identifying an individual of the same object as an attribute of the object. In the object identification device of claim 1,
The object identification device, wherein the movement line calculation unit calculates a plurality of movement lines for each part of an object or for each feature point in an object region. In the object identification device of claim 1,
An object identification apparatus identifies the type of an object or the attribute of an object by analyzing the shape of an object in addition to the flow line obtained by the flow line calculation unit. In the object identification device of claim 1,
An object is detected from a plurality of moving images acquired from a plurality of imaging devices, a flow line of the object in each moving image is determined, and the shape of the flow line is analyzed. An object identification apparatus that identifies whether or not there is an object. In the object identification device of claim 1,
An object identification apparatus characterized by identifying at least one of gender, age, clothes, and the type of package carried as an attribute of the object. In the object identification device of claim 1,
An object identification apparatus characterized in that a length of a flow line used for analysis of a shape is determined from a flow line obtained by a flow line calculation unit based on a cycle of movement operation of an object. In the object identification device according to claim 1,
An object identification apparatus , wherein the flow line is a series of center of gravity of a detected object . An object detection unit that detects an object moving from a moving image;
A movement line calculation unit for calculating a movement line of the object detected by the object detection unit;
An object identification unit that identifies the type of the object or the attribute of the object based on the shape of the flow line obtained by the flow line calculation unit;
An object identification apparatus, which calculates the reliability of the identification result of the type or the attribute based on the length of the flow line obtained by the flow line calculation unit . An object detection unit that detects an object moving from a moving image;
A movement line calculation unit for calculating a movement line of the object detected by the object detection unit;
An object identification unit that identifies the type of the object or the attribute of the object based on the shape of the flow line obtained by the flow line calculation unit;
An image processing apparatus characterized by calculating the reliability of the identification result of the type or the attribute based on the length of the flow line obtained by the flow line calculation unit .