JP2013153842A - Detector, detection method, and detection program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a detector capable of accurately detecting movements of a patient by simple processing.SOLUTION: A detector receives input of two or more photographed images obtained by continuously photographing a subject from a camera (#1), and extracts a feature point from each of a first photographed image and a second photographed image (#2). The feature point of the first photographed image and the feature point of the second photographed image that correspond are matched to detect a moving direction for each feature point, the moving directions are classified for each corresponding feature point, and the movement of the subject is detected on the basis of the number of feature points for each moving direction, that is obtained by subtracting the number of feature points in the moving direction which is a direction opposite to the moving direction from the number of feature points for each moving direction (#3). On the basis of the detected movement of the subject, whether or not to output information associated with the movement is determined (#4).

Description

  The present invention relates to a detection device, a detection method, and a detection program, and more particularly, to a detection device, a detection method, and a detection program for detecting a motion of a patient.

  Various methods have been proposed for detecting the fall of a patient in a facility, for example, a patient on a bed or walking in a corridor. For example, Japanese Patent Application Laid-Open No. 2002-232870 (hereinafter, Patent Document 1) discloses a detection device that detects the fall of a cared person from image data of the cared person.

  The detection device disclosed in Patent Document 1 calculates a motion vector of a cared person from image data obtained by photographing the cared person, and uses a predetermined threshold T indicating the fall of the cared person. By comparing, it is determined whether or not the care recipient has fallen. Therefore, it is possible to quickly report the caregiver's fall.

JP 2002-232870 A

  By the way, the motion vector is obtained by calculating a difference between frames of image data, that is, an optical flow. For this reason, when the image data of the cared person is affected by noise due to the surrounding environment and brightness, the motion of the patient is not accurately represented by the motion vector. As a result, there is a problem in that the detection accuracy decreases when an operation such as a patient's falling is detected using the motion vector.

  Further, there are cases where various parts of the patient make various movements for one operation. For example, when moving up from the bed, the shoulder may move upward, but the hand may move downward to support the movement away from the bed. Thus, since each part may move in various directions even in one operation, the motion vector is detected not only in a fixed direction but also in various directions. For this reason, there is a problem that the detection accuracy decreases when an operation such as a patient's falling is detected using such an operation vector.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a detection device, a detection method, and a detection program that can accurately detect a patient's motion with simple processing.

  In order to achieve the above object, according to one aspect of the present invention, the detection device is a detection device that detects a motion of a subject, and inputs two or more captured images obtained by continuously photographing the subject. Input means for processing, processing means for performing processing to detect the motion of the subject by analyzing the captured image, and output means for outputting information related to the motion of the subject detected by the processing means Prepare. The processing means includes extraction means for extracting feature points from each of the first photographed image and the second photographed image obtained in succession, the feature points of the corresponding first photographed image, and the second feature points. The movement direction for each corresponding feature point is detected by comparing the feature points of the captured image, the movement direction is classified for each corresponding feature point, and the direction opposite to the movement direction is determined from the number of feature points for each movement direction. Detection means for detecting the movement of the subject based on the number of feature points for each movement direction obtained by subtracting the number of feature points in the movement direction, and the movement of the subject detected by the output means And determining means for determining whether or not to output information related to the operation.

  Preferably, the extracting means compares the brightness of the corresponding points of the first captured image and the second captured image, and uses the point where the difference is equal to or greater than a prestored threshold as a feature point. Extract.

  Preferably, in the detection unit, the movement distance between the corresponding feature point extracted from the first photographed image and the feature point extracted from the second photographed image is equal to or greater than a threshold value stored in advance. For the feature points, the movement direction of the feature points is classified into predetermined movement direction sections.

  Preferably, the detection means detects the moving direction having the largest number of feature points for each classified moving direction as the moving direction of the subject.

  Preferably, the detection means calculates the sum of the vectors for each movement direction, using the number of feature points for each classified movement direction as the length of the vector for the movement direction, and represents the sum of the calculated vectors. The direction passed is detected as the moving direction of the subject.

  Preferably, the detection unit divides each of the first photographed image and the second photographed image into predetermined regions, and sets the direction opposite to the moving direction from the number of feature points in each moving direction for each region. Based on the number of feature points for each movement direction obtained by subtracting the number of feature points in the movement direction, the movement direction in the area is specified, and the movement direction of the subject is detected based on the number of areas for each movement direction. .

  More preferably, the detection means detects the movement direction having the largest number of regions for each classified movement direction as the movement direction of the subject.

  According to another aspect of the present invention, the detection method is a method for detecting an operation of a subject from a photographed image by a detection device, and the step of inputting two or more photographed images obtained by continuously photographing the subject. Extracting a feature point from each of the first captured image and the second captured image obtained in succession, and a corresponding feature point of the first captured image and a feature point of the second captured image The movement direction for each corresponding feature point is detected by comparing with the above, and the movement direction is classified for each corresponding feature point, and the direction is opposite to the movement direction from the number of feature points for each movement direction. Detecting the motion of the subject based on the number of feature points in each movement direction obtained by subtracting the number of feature points in the movement direction.

  According to still another aspect of the present invention, the detection program is a program for causing the detection device to execute processing for detecting the motion of the subject from the captured image, and two or more obtained by continuously capturing the subject. A step of inputting the captured image, a step of extracting feature points from each of the first captured image and the second captured image obtained in succession, and a corresponding feature point of the first captured image Detecting the moving direction for each corresponding feature point by comparing the feature points of the two captured images, classifying the moving direction for each corresponding feature point, and moving from the number of feature points for each moving direction And detecting a motion of the subject based on the number of feature points in each movement direction obtained by subtracting the number of feature points in the movement direction that is opposite to the direction.

  According to the present invention, it is possible to detect a patient's movement with high accuracy by simple processing.

It is a figure which shows the specific example of a structure of the safety nursing system concerning embodiment. It is a figure for demonstrating an example of the attachment position of the camera contained in a safety nursing system. It is a block diagram which shows the specific example of a structure of the detection apparatus contained in a safety nursing system. It is a figure showing the detection method in a detection apparatus. It is a block diagram which shows the specific example of a function structure of a detection apparatus. It is a flowchart showing the flow of operation | movement in a detection apparatus. It is the figure showing the classification result of the moving direction for every moving point, and the division of the moving direction obtained by the inventors' measurement. It is a figure for demonstrating the modification of a detection method.

  Embodiments of the present invention will be described below with reference to the drawings. In the following description, the same parts and components are denoted by the same reference numerals. Their names and functions are also the same.

<System configuration>
FIG. 1 is a diagram showing a specific example of the configuration of the safety nursing system 1 according to the present embodiment.

  Referring to FIG. 1, safety nursing system 1 includes a camera 151 and a detection device 100 connected to camera 151.

  FIG. 2 is a diagram for explaining an example of the attachment position of the camera 151. With reference to FIG. 2, the camera 151 of the safety nursing system 1 is arrange | positioned in the position which can image the patient 10 on the bed 20 from a head direction as an example.

  In addition, as long as the camera 151 is arrange | positioned in the position which can image including the side edge 23 of the upper surface of the bed 20, it is not limited to the position which can image from the patient's 10 head direction, The foot part of the patient 10 You may arrange | position in the position which can image from a direction.

  Further, the camera 151 is preferably arranged at a position where imaging is possible including the longitudinal end 24 on the upper surface of the bed 20.

  In addition, the patient 10 is a person who needs to be watched by others such as a nurse who receives nursing care from a nurse such as a doctor and a nurse at a hospital, and a care recipient who receives care from a caregiver at a care facility. Including.

<Device configuration>
FIG. 3 is a block diagram illustrating a specific example of the configuration of the detection apparatus 100.

  Referring to FIG. 3, detection device 100 includes a control unit 110, a storage unit 120, an operation unit 130, a display unit 140, an image input unit 150, an audio output unit 160, and a communication unit 190. .

  The control unit 110 includes a CPU (Central Processing Unit) and its auxiliary circuit, and controls each unit of the detection apparatus 100. That is, by executing predetermined processing according to the program and data stored in the storage unit 120, the data input from the operation unit 130, the image input unit 150, and the communication unit 190 is processed, and the processed data is The data is stored in the storage unit 120, is output as audio by the audio output unit 160, or is output from the communication unit 190.

  The storage unit 120 includes a RAM (Random Access Memory) used as a work area necessary for executing the program by the control unit 110 and a ROM (Read) for storing a basic program to be executed by the control unit 110. Only Memory). Further, as a storage medium of an auxiliary storage device for assisting the storage area of the storage unit 120, a magnetic disk (HD (Hard Disk), FD (Flexible Disk)), optical disk (CD (Compact Disc), DVD (Digital Versatile Disk) ), BD (Blu-ray Disc)), a magneto-optical disk (MO (Magneto-Optical disk)), or a semiconductor memory (memory card, SSD (Solid State Drive)) may be used.

  The operation unit 130 includes a keyboard and a mouse, and transmits an operation signal indicating an operation by the user to the control unit 110. Further, the operation unit 130 may be configured by another operation device such as a touch panel instead of or in addition to the keyboard and the mouse.

  Display unit 140 includes a display (for example, LCD (Liquid Crystal Display)) and the like. The display unit 140 is controlled by the control unit 110 to display a predetermined image on the display.

  The audio output unit 160 includes a speaker. The audio output unit 160 is controlled by the control unit 110 to output predetermined audio from the speaker.

  The image input unit 150 transfers a captured image input from the camera 151 to the control unit 110, or controls the control unit 110 to store the captured image in the storage unit 120 or display the image on the display unit 140.

  The communication unit 190 transmits information from the control unit 110 to another device, receives information transmitted from the other device, and passes the information to the control unit 110. Examples of other devices include a biological information monitor (not shown) and a data storage server.

  3 shows an example in which the operation unit 130, the display unit 140, the image input unit 150, and the audio output unit 160 are all directly connected to the main body of the detection apparatus 100. May be remotely installed and connected via the communication unit 190.

  For example, the detection device 100 is installed in the vicinity of the camera 151, that is, installed in a hospital room provided with the bed 20, and the operation unit 130, the display unit 140, the image input unit 150, and the audio output unit 160 are installed in the nurse station. The communication unit 190 may be used for communication.

  As another form, only the camera 151 is installed in a hospital room provided with the bed 20, and the main body of the detection apparatus 100 including the operation unit 130, the display unit 140, the image input unit 150, and the audio output unit 160 is a nurse station. For example, the image information from the camera 151 may be input to the detection apparatus 100 via the communication unit 190.

<Overview of operation>
The safety nursing system 1 captures a patient on a bed, who is a care recipient, continuously with the camera 151 as shown in FIG. 2, and analyzes the captured image with the detection device 100, thereby detecting the detection target of the patient. Detect that there was an action.

FIG. 4 is a diagram illustrating a detection method in the detection apparatus 100.
Referring to FIG. 4, detection apparatus 100 accepts input of a plurality of captured images that are continuously captured from camera 151 (# 1).

  A feature point that is a moving point is extracted from the previous captured image by comparing the previous captured images of the input captured images arranged in time series (# 2). Specifically, the brightness of each of these two captured images is compared, and a point whose difference is equal to or greater than a prescribed threshold value is extracted as a feature point.

  By matching the feature points extracted from the input photographed image with the feature points of the previous photographed image arranged in time series, the moving direction for each feature point is specified (# 3). Specifically, a distance between corresponding feature points in two captured images is calculated, a point whose distance is equal to or greater than a prescribed threshold is specified as a moving point, and a moving direction thereof is specified. The “moving point” can be said to be a point on the photographed image representing the moved part.

The movement of the care recipient is determined based on the movement direction for each feature point (# 4).
The detection apparatus 100 outputs a warning using the display unit 140, the audio output unit 160, and the like according to the determination result.

<Functional configuration>
FIG. 5 is a block diagram illustrating a specific example of a functional configuration of the detection apparatus 100 for realizing the above-described operation. Each function of FIG. 5 is mainly formed in the CPU by reading and executing a program stored in the storage unit 120 by a CPU (not shown) included in the control unit 110. At least a part of the functions shown in FIG. 3 may be realized by the apparatus configuration shown in FIG.

  Referring to FIG. 5, detection apparatus 100 temporarily stores input unit 101 for receiving input of a captured image from image input unit 150, the input captured image, and data that is being processed later. For comparing the brightness of each of the corresponding points between the temporary storage unit 102 for the input and the captured image input immediately before and the captured image input immediately before, and the difference is equal to or greater than a threshold value stored in advance A first extraction unit 103 for extracting a point on the captured image as a feature point, a feature point extracted from the input captured image, and a feature point extracted from the captured image input immediately before A second extraction unit 104 for matching and extracting a feature point on the input photographed image whose difference in position is equal to or greater than a pre-stored threshold value as a movement point; and for each extracted movement point Entered just before A classifying unit 105 for specifying the moving direction from the feature point extracted from the photographed image and sorting the moving direction segment stored in advance, and the moving classified into the paired segment whose moving direction is the opposite direction By canceling the number of points, the detection unit 106 for detecting the movement direction as a whole and determining whether or not the detected movement direction corresponds to the movement direction of the warning object stored in advance. Thus, the determination unit 107 for determining whether or not the care recipient has performed the warning target operation, and the output unit 108 for performing a process of outputting a warning from the voice output unit 160 or the like according to the determination result, for example. Including.

<Operation flow>
FIG. 6 is a flowchart showing an operation flow in the detection apparatus 100. The operation shown in the flowchart of FIG. 6 is performed by a CPU (not shown) included in the control unit 110 of the detection apparatus 100 reading out and executing a program stored in the storage unit 120 and causing each unit shown in FIG. 5 to function. This is realized.

  Referring to FIG. 6, in step S101, the CPU accepts an input of a captured image from image input unit 150, and extracts a feature point from the image in step S103. In step S103, a difference in luminance between corresponding points between the captured image (t) input in step S101 and the captured image (t-1) input in the previous process is calculated. Then, the difference is compared with a threshold value stored in advance, and a point on the captured image (t) for which a difference equal to or greater than the threshold is calculated is extracted as a feature point.

  Next, in step S105, the CPU extracts a movement point from the feature points from the captured image (t) extracted in step S103. In step S105, the feature point extracted from the photographed image (t) in step S103 is matched with the feature point extracted from the photographed image (t-1) in the previous process. Then, the position difference is compared with a threshold value stored in advance, and a point on the captured image (t) having a position difference equal to or greater than the threshold value is extracted as a moving point.

  In step S107, the CPU classifies the movement direction of each movement point extracted from the captured image (t) in step S107 based on the movement direction from the feature point of the captured image (t-1). Here, for example, divisions such as “up”, “down”, “right”, and “left” are set in advance, and the movement direction of each movement point is compared with the range of the direction of each division. This classifies which of these categories corresponds to the category.

  In step S109, the movement direction of the care receiver as a whole is detected according to the number of movement points for each category. Here, in the above-described example, the number of moving points classified into each of the paired sections in which the moving directions are opposite directions, which are “upper” and “lower”, “right” and “left”, is canceled. That is, the movement direction of “up” or “down” is detected by canceling out the number of movement points classified as “upper” and the number of movement points classified as “lower”. Similarly, the movement direction of “right” or “left” is detected by canceling out the number of movement points classified as “right” and the number of movement points classified as “left”.

  The CPU stores the moving direction of the warning target in advance. For example, when the cared person is a patient on a bed, the movement direction of the patient is stored as the movement direction of the warning object by storing the movement direction of the body when the patient is rising as the movement direction of the warning object. Is detected. The CPU outputs a warning in step S113 when the movement direction detected in step S109 matches the stored movement direction of the warning object or is within a predetermined range (YES in step S111). . And a series of operation | movement is complete | finished.

  Since the captured images from the camera 151 are input continuously, the series of operations shown in FIG. 6 is repeated each time the captured image from the camera 151 is input.

<Effect of Embodiment>
By performing the above operation, the moving direction of the care recipient is detected by reflecting the moving direction of each part in the entire moving direction. Therefore, for example, when moving up from the bed, the shoulders move upward, but the hand moves downward to support the movement away from the bed. Even in this case, since the entire movement direction is captured from each movement direction, the movement direction of the care receiver as a whole can be detected with high accuracy by simple processing.

  Furthermore, the detection device 100 detects a moving direction by extracting feature points based on the brightness of the captured image and comparing the feature points. That is, since the detection apparatus 100 does not use the captured image itself as a detection result, the movement direction can be detected without impairing the privacy of the patient who is the care recipient.

  In this way, by detecting the moving direction of the care recipient with high accuracy, the moving direction of the motion that becomes a precursor of the motion to be avoided, such as falling from the bed or falling in the hallway, is set as the moving direction of the warning target. Therefore, it is possible to detect the precursor of these operations, and to suppress the occurrence of such operations. That is, for a fall from the bed, the movement direction of the movement of the patient, who is the care receiver, getting up from the bed is set as the warning object movement direction. As a result, it is possible to detect rising of the care receiver from the bed, and it is possible to avoid a subsequent fall from the bed by dealing with the warning at that stage.

<Measurement example>
Note that the inventors have actually performed measurements to detect the direction of movement of the care recipient using the safety nursing system 1.

  FIG. 7A is a diagram showing the classification result of the moving direction for each moving point obtained by the inventors' measurement. In this measurement, the camera 151 is installed in the direction from the head of the cared person to the foot while getting up from the prone state on the bed (in the direction of photographing the back of the cared person), and continuous shooting is performed. A captured image (t-1) and a captured image (t) were obtained.

  There are 261 moving points obtained from the photographed image (t), and these are shown as “right”, “lower right”, “lower left”, “left”, “upper left”, “ "Upper right". As a result, from the captured image (t), the moving point classified as “right” is classified as 21.1% of the whole, the moving point classified as “upper right” is classified as 22.2% of the whole, and “upper left”. The transferred points accounted for 21.1% of the total. In addition, the moving point classified as “left” is 13.4% of the whole, the moving point classified as “lower left” is 10.0% of the whole, and the moving point classified as “lower right” is the whole. It was 12.3%.

  If the conventional method of detecting the motion direction using the motion vector obtained by calculating the difference between frames of the captured image is used, in the example of FIG. 7, “right”, “upper right”, and “upper left” Since the moving points classified as "" are large at the same rate, it is unclear which direction the movement is the largest.

  On the other hand, the detection apparatus 100 detects the movement direction in the opposite direction by canceling the number of movement points in the opposite movement direction. That is, in the example of FIG. 7, “Right” is 19.6% by canceling the ratio of “Right” and “Left”, and “Right” is canceled by canceling the ratio of “Upper right” and “Lower left”. Was 57.4%, and “upper left” was 22.8% by offsetting the ratio of “upper left” and “lower right”. In other words, it was detected by simple processing that the ratio of “upper right” was the highest. Thereby, as represented by the arrow in FIG. 7, it was detected that the whole moved in the upper right direction.

[Modification 1]
In the above example, as shown in FIG. 7, for example, the number of moving points for each moving direction section is offset with the number of moving points in the opposite moving direction in the entire image.

  However, for example, as shown in FIG. 8, the input image is divided into predetermined areas, and the number of movement points for each of the movement direction sections in the divided areas is set in the opposite direction. The direction of movement in the region is determined by canceling the number of movement points in the movement direction, and then the number of movement points in each movement direction section is determined as the number of movement points in the opposite direction in the entire image. You may make it cancel.

  Alternatively, in this case, for each divided small area, the movement direction having the largest number of movement points for each movement direction section in the area is set as the movement direction of the area, and then the small movement for each movement direction section is performed. You may make it cancel the number of area | regions with the number of the small areas of the moving direction of the opposite direction in the whole image.

  In this way, for example, the influence of the movement direction obtained at the movement point due to movements other than the care recipient, such as wrinkles on clothes, wrinkles on sheets, and curtains, can be suppressed and the detection accuracy can be improved. Can be increased.

[Modification 2]
In the above example, after canceling out the number of movement points in the opposite movement direction, the movement direction with the largest number of movement points is assumed to be the overall movement direction. In the example of FIG. 7, since the ratio of the moving points divided into “upper right” is the highest, it is detected that the moving point is moving in the upper right direction as a whole.

  As another example, a vector corresponding to the direction of each section is specified according to the number of moving points for each section after canceling with the number of moving points in the opposite direction of movement, and the sum of these vectors is obtained. The direction may be the entire moving direction. In the example of FIG. 7, for example, the length of the “right” vector is 19.6, the length of the “upper right” vector is 57.4, and the length of the “upper left” vector is 22.8. The direction of the vector obtained by the sum of the vectors may be the overall movement direction.

By doing in this way, the whole moving direction can be detected with high accuracy.
[Modification 3]
In the above example, it is assumed that the shooting interval is short to some extent, or the movement of the care recipient is sufficiently slow with respect to the shooting interval, and the shot image (t) is included in the shot image (t−1). It is assumed that the portion corresponding to the target portion in the center is at the same position as the captured image (t).

  However, when the shooting interval is long or the movement of the care recipient is faster than the shooting interval, the portion corresponding to the target portion in the shot image (t) is not in the shot image (t-1) or In some cases, the position is greatly changed and difficult to catch.

  In such a case, the detection apparatus 100 calculates the distribution range of the feature point in the photographed image after performing step S103 to extract the feature point in the photographed image (t). Then, by comparing with the distribution range in the photographed image (t-1), the change direction of the distribution range can be detected as the overall movement direction of the care receiver.

  For example, when the patient who is a care recipient gets up from the bed as an example, in the photographed image by the camera 151 installed on the head, first, after the top of the head is photographed, the entire head, From the shoulders to the shoulders, the upper body, and the feature points in the captured image spread from the bottom to the entire image. In that case, since it is detected that the distribution range of the feature point in the captured image moves upward, the upward direction can be detected as the movement direction of the entire care receiver.

  In this way, the entire moving direction can be detected with high accuracy by a simple process.

  Further, it can be provided as a program for causing the detection apparatus 100 to execute the above-described detection processing. Such a program is stored in a computer-readable recording medium such as a flexible disk attached to the computer, a CD-ROM (Compact Disk-Read Only Memory), a ROM (Read Only Memory), a RAM (Random Access Memory), and a memory card. And can be provided as a program product. Alternatively, the program can be provided by being recorded on a recording medium such as a hard disk built in the computer. A program can also be provided by downloading via a network.

  The program according to the present invention is a program module that is provided as a part of a computer operating system (OS) and calls necessary modules in a predetermined arrangement at a predetermined timing to execute processing. Also good. In that case, the program itself does not include the module, and the process is executed in cooperation with the OS. A program that does not include such a module can also be included in the program according to the present invention.

  The program according to the present invention may be provided by being incorporated in a part of another program. Even in this case, the program itself does not include the module included in the other program, and the process is executed in cooperation with the other program. Such a program incorporated in another program can also be included in the program according to the present invention.

  The provided program product is installed in a program storage unit such as a hard disk and executed. The program product includes the program itself and a recording medium on which the program is recorded.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

  DESCRIPTION OF SYMBOLS 1 Safety nursing system, 10 patients, 20 beds, 23 side edge, 24 longitudinal direction edge, 100 detection apparatus, 101 input part, 102 temporary storage part, 103 1st extraction part, 104 2nd extraction part, 105 classification part, 106 Detection unit, 107 determination unit, 108 output unit, 110 control unit, 120 storage unit, 130 operation unit, 140 display unit, 150 image input unit, 151 camera, 160 audio output unit, 190 communication unit.

Claims (9)

A detection device for detecting the movement of a subject,
An input means for inputting two or more captured images obtained by continuously capturing the subject;
Processing means for analyzing the captured image to detect a motion of the subject;
Output means for outputting information related to the motion of the subject detected by the processing means;
The processing means includes
Extraction means for extracting feature points from each of the first captured image and the second captured image obtained successively;
By comparing the corresponding feature point of the first photographed image with the feature point of the second photographed image, the movement direction for each corresponding feature point is detected, and the movement direction is determined for each corresponding feature point. The subject based on the number of feature points for each movement direction obtained by classifying and subtracting the number of feature points for the movement direction opposite to the movement direction from the number of feature points for each movement direction Detection means for detecting the operation of
A detection apparatus further comprising: a determination unit for determining whether to output information related to the operation based on the operation of the subject detected by the detection unit.   The extraction means compares the brightness of each corresponding point between the first captured image and the second captured image, and the feature point is that the difference is equal to or greater than a prestored threshold value. The detection device according to claim 1, which is extracted as   In the detection means, the movement distance between the corresponding feature point extracted from the first photographed image and the feature point extracted from the second photographed image is equal to or greater than a threshold value stored in advance. The detection device according to claim 1, wherein the moving direction of the feature point is classified into predetermined moving direction sections for the feature point.   The detection device according to claim 1, wherein the detection unit detects a movement direction having the largest number of feature points for each classified movement direction as the movement direction of the subject.   The detection means calculates the sum of the vectors for each movement direction, using the number of feature points for each classified movement direction as the length of the vector in the movement direction, and represents the sum of the calculated vectors. The detection apparatus according to claim 1, wherein the detected direction is detected as a moving direction of the subject.   The detection means divides each of the first photographed image and the second photographed image into predetermined areas, and in each area, the number of feature points for each movement direction is opposite to the movement direction. The movement direction in the region is specified based on the number of feature points in each movement direction obtained by subtracting the number of feature points in the movement direction, and the subject is determined based on the number of regions in each movement direction. The detection device according to claim 1, wherein the detection device is detected as a moving direction.   The detection device according to claim 6, wherein the detection unit detects a movement direction having the largest number of the regions for each classified movement direction as a movement direction of the subject. A method for detecting a motion of a subject from a captured image by a detection device,
Inputting two or more captured images obtained by continuously capturing the subject;
Extracting feature points from each of the first captured image and the second captured image obtained successively;
Detecting a moving direction for each corresponding feature point by comparing the corresponding feature point of the first captured image with the feature point of the second captured image;
For each movement direction obtained by classifying the movement direction for each corresponding feature point and subtracting the number of feature points in the movement direction opposite to the movement direction from the number of feature points for each movement direction. Detecting the motion of the subject based on the number of feature points. A program for causing a detection device to execute processing for detecting a motion of a subject from a captured image,
Inputting two or more captured images obtained by continuously capturing the subject;
Extracting feature points from each of the first captured image and the second captured image obtained successively;
Detecting a moving direction for each corresponding feature point by comparing the corresponding feature point of the first captured image with the feature point of the second captured image;
For each movement direction obtained by classifying the movement direction for each corresponding feature point and subtracting the number of feature points in the movement direction opposite to the movement direction from the number of feature points for each movement direction. And a step of detecting the motion of the subject based on the number of feature points.

Images