JP4621171B2 - Feature point movement detection apparatus and feature point movement detection method - Google Patents

Description

  The present invention relates to an apparatus and a method for detecting a movement amount of a feature point existing in data, for example, from data such as a plurality of images taken continuously.

  As a method for detecting the motion of an object in the screen from two images that are shifted in time, a method for obtaining corresponding points in a plurality of images and detecting the motion, or a method called “optical flow” is used. It has been. In the method of detecting the movement of the feature point from the corresponding point, the calculation process for obtaining the corresponding point is complicated. The optical flow has advantages in that the calculation process is easier than the method for obtaining the corresponding points, and information on the velocity of the object can be obtained.

  However, since the optical flow cannot determine the correspondence between points, it has an essential problem that it must be regarded as a point-to-point correspondence by looking at the coincidence between the regions.

In addition to the above, a motion detection device described in Patent Document 1 is known as a method for detecting the motion of an object in a screen from a time-series image. The motion detection device of Patent Document 1 detects an edge from an image and counts up a count value for a pixel in which the edge exists. Thereby, since the count value indicates the movement state of the edge, the movement direction and movement speed of the edge can be obtained based on the count value of each pixel. This method can provide continuity by inserting a plurality of points between edges detected in a temporally shifted image, and can track the correspondence between the points.
JP 2006-92427 A

  However, in the above-described motion detection device of Patent Document 1, the count value is incremented as long as an edge is detected. Therefore, when the moving speed of the object is slow, the same pixel is repeatedly counted up and the memory overflows. End up. In Patent Document 1, in order to solve this problem, the concept of “vanishing point” is introduced, the inside of an image is divided into a plurality of regions based on the vanishing point, and the sampling rate is changed according to the region.

  The above-described method for dividing an image into a plurality of regions requires processing at different sampling rates depending on the region, and thus the calculation processing becomes complicated. In this method, in a region where the movement of the object is slow, the sampling rate is reduced to make it difficult for the memory to overflow. However, the movement of all the objects in the image is gradually reduced toward the vanishing point. It is not a complete solution.

  In view of the above background, an object of the present invention is to provide a feature point movement detection device that can detect the movement of feature points included in data input by simple processing.

  The feature point movement detection device of the present invention is a device for detecting feature point movement, a feature point detection unit for detecting feature points from input data, and a feature point detected by the feature point detection unit. A straight line calculation unit for obtaining a straight line having an inclination determined based on the timing at which the feature point is detected, and a straight line calculated by the straight line calculation unit for the feature point detected at a different timing are stored. A plurality of lines detected at different timings based on a crossing state of a plurality of straight lines stored in the memory, a write control unit that writes the straight lines calculated by the straight line calculation unit to the memory, A movement detection unit that obtains correspondence between the feature points and detects movement of the feature points.

  In this way, feature points that move with the passage of time are detected at a predetermined timing, and a straight line having an inclination determined based on the detection timing is drawn for a plurality of feature points detected at different timings. Assuming that the moving speed of the feature points is constant, the straight line group intersects at a certain point. In the present invention, the movement of the feature point can be detected by a simple process of obtaining the correspondence of the feature point obtained at different detection timings based on the intersection state of the straight lines passing through the feature point. In addition, as the moving speed of the feature point increases, the distance from the feature point to the intersection of the straight lines increases. Therefore, the moving speed of the feature point can be expressed by the distance from the feature point to the intersection.

  In the feature point movement detection device, the memory prepares a counter corresponding to each coordinate of a predetermined coordinate system that defines the straight line, and the writing control unit is calculated by the straight line calculation unit. The counter corresponding to the coordinates of each point constituting the straight line may be counted up, and the movement detection unit may obtain the intersection state of the straight lines based on the value of the counter corresponding to each coordinate of the memory.

  In this way, by counting up the counters corresponding to the points constituting each straight line, the value of the counter becomes large at the place where the straight lines intersect, and the counter having a value of 1 or more around the place where the straight lines intersect. Density increases. Therefore, it is possible to easily grasp the intersection state of the straight lines based on the counter value. Moreover, since the coordinates of the peak of the counter value are directly used as information regarding the movement of the feature point, the movement of the feature point can be detected without requiring complicated calculation.

  In the feature point movement detection device, the memory has a plurality of areas, and the write control unit starts writing data to each of the plurality of areas from different timings. The movement of the feature points may be detected with reference to an area in which the number of data writing reaches a predetermined number among the plurality of areas.

  Thus, by writing data to each of the plurality of regions of the memory from different timings, it is possible to detect the movement of the feature points based on the data of any region.

  In the feature point movement detection device, the writing control unit may erase data in a region used by the movement detection unit to detect movement of the feature point.

  As described above, by deleting the data of the region used for detecting the movement of the feature point, it is possible to write the feature point newly detected using the region.

  In the feature point movement detection device, the straight line calculation unit may set the slope θ of the straight line so that the change in tan θ is proportional to the interval of the detection timing of the feature points.

  Thus, by making the change in tan θ proportional to the feature point detection time interval, each straight line passing through the feature point moving at a constant movement speed intersects at a predetermined point.

  The feature point movement detection method of the present invention is a method for detecting feature point movement, a feature point detection step for detecting a feature point from input data, and a feature point detected by the feature point detection step. And a straight line calculation step for obtaining a straight line having an inclination determined based on the timing at which the feature point is detected, and a straight line calculated in the straight line calculation step for feature points detected at different timings are stored. Detecting at different timings based on a step of preparing a memory to be stored, a writing control step of writing the straight line calculated in the straight line calculating step into the memory, and an intersection state of a plurality of straight lines stored in the memory A movement detection step of obtaining correspondence between the plurality of feature points and detecting movement of the feature points.

  With this configuration, as with the feature point movement detection device of the present invention, a complicated calculation is performed by a simple process of obtaining a straight line of an inclination determined based on the detection timing for a feature point detected from data. It is possible to detect the movement of feature points without processing. In addition, various configurations of the feature point movement detection apparatus of the present invention can be applied to the feature point movement detection method of the present invention.

  The program of the present invention is a program for detecting movement of a feature point, and a feature point detection step for detecting a feature point from data input to a computer, and a feature detected in the feature point detection step. A straight line calculation step for obtaining a straight line that passes through a point and has an inclination determined based on the timing at which the feature point is detected, and a straight line calculated in the straight line calculation step for the characteristic point detected at a different timing. Based on a step of generating a memory area to be stored, a writing control step of writing the straight line calculated in the straight line calculating step into the memory area, and an intersection state of a plurality of straight lines stored in the memory area, A movement detecting step for obtaining correspondence between the plurality of feature points detected at different timings and detecting movement of the feature points; Make.

  With this configuration, as with the feature point movement detection device of the present invention, a complicated calculation is performed by a simple process of obtaining a straight line of an inclination determined based on the detection timing for a feature point detected from data. It is possible to detect the movement of feature points without processing. In addition, various configurations of the feature point movement detection device of the present invention can be applied to the program of the present invention.

  According to the present invention, it is possible to detect the movement of a feature point by a simple process of obtaining a straight line of an inclination determined based on the detection timing for a feature point detected from data. .

  Hereinafter, a feature point movement detection apparatus according to an embodiment of the present invention will be described. In the following, an apparatus for detecting movement of a car shown in an image will be described as an example. However, the feature point movement detection device of the present invention is not limited to a device that detects the movement of feature points included in an image. For example, the feature point movement detection device can also detect the movement of the detected feature point from data obtained by an infrared sensor or the like.

  FIG. 1 is a diagram illustrating a configuration of a feature point movement detection device 1 according to the present embodiment. The feature point movement detection apparatus 1 includes an image input unit 10 that inputs an image, an edge image generation unit 12 that extracts a vertical edge from the input image and generates a vertical edge image, and a feature from the generated vertical edge image. And a feature point detector 14 for detecting points.

  FIG. 2 is a diagram illustrating an example of a vertical edge image generated by the edge image generation unit 12. In the present embodiment, the movement of the automobile in the horizontal direction is detected. The feature point detection unit 14 reads pixel data of one line in the horizontal direction from the vertical edge image generated by the edge image generation unit 12, and detects feature points from the read pixel data. In the present embodiment, since the vertical edge image is used, the movement of the feature point can be detected with high accuracy when the feature point moves in the horizontal direction. When detecting the movement of the feature point in the vertical direction, a configuration is adopted in which a horizontal edge image is generated and pixel data is read in the vertical direction.

  The feature point movement detection device 1 includes a feature point coordinate calculation unit 16 that calculates the coordinates of the feature points detected by the feature point detection unit 14, and a feature point coordinate as a configuration for associating a plurality of detected feature points. Using the position coordinates calculated by the calculation unit 16, a straight line calculation unit 18 that obtains a straight line passing through the point and a memory 24 that stores the obtained straight line are provided. The straight line calculation unit 18 reads the offset value corresponding to the straight line inclination from the inclination table 20 and generates straight line data.

  The feature point coordinate calculation unit 16 calculates the coordinates of the feature points detected by the feature point detection unit 14 in a predetermined coordinate system. Here, since the movement of the feature point in the horizontal direction is detected, the coordinates of the feature point are obtained in the form of (x, 0).

  Here, the principle of feature point movement detection of the present embodiment will be described using the example shown in FIG. In the example illustrated in FIG. 2, the feature point a and the feature point b are detected by the feature point detection unit 14.

  FIG. 3 is a diagram illustrating movement of feature points. As shown in FIG. 2, since the automobile is traveling in the right direction, the feature point a and the feature point b move to the right as time passes. In FIG. 3, the feature points detected at time t1 are points c1 and c2, the feature points detected at time t2 are points c3 and c4, and the feature points detected at times t3 to t5 are points c5 to c10. It is described.

  At the stage where the feature points are detected from the image by the feature point detector 14, the correspondence between the points c1 to c10 detected at each detection time is not known. For example, the feature point c1 at the time t1 may move to the position of the feature point c4 at the time t2, and the feature point c2 at the time t2 may be a newly appearing feature point. May disappear from the predetermined coordinate system, and the point c2 at time t1 may correspond to the point c3 at time t2. The feature point movement detection device 1 of the present embodiment associates the points c1 to c10 detected at each time with a simple process.

  FIG. 4 is a diagram for explaining processing performed by the feature point movement detection apparatus 1. In the present embodiment, the moving speed of the feature points is constant, and the feature point movement detection apparatus 1 detects the feature points at a constant time interval. Therefore, the feature points detected by the feature point movement detection device 1 are equally spaced. Note that if feature points are detected at sufficiently small time intervals, the moving speed of the feature points can be considered to be substantially constant, so that it can be assumed that the moving speed of the feature points is constant as described above.

  The feature point movement detection device 1 draws a straight line passing through the detected points c1 to c10. At this time, the inclination of the straight line is determined according to the timing at which the points c1 to c10 are detected. Lines l1 and l2 having a slope θ1 are drawn at points c1 and c2 detected at time t1. Lines l3 and l4 having a slope θ2 are drawn at points c3 and c4 detected at time t2. Similarly, straight lines l5 and l6 are drawn at points c5 and c6, straight lines l7 and l8 are drawn at points c7 and c8, and straight lines l9 and l10 are drawn at points c9 and c10.

  Here, the inclination θ of the straight line is determined so that the change in tan θ is constant. In the example shown in FIG. 4, tan (θ1) −tan (θ2) = tan (θ2) −tan (0) = tan (0) −tan (−θ2) = tan (−θ2) −tan (−θ1) A relationship is established. When a straight line as described above is drawn, straight lines passing through feature points acquired at a plurality of times at the same time interval intersect at predetermined points. In the present embodiment, it is determined that the feature points included in each of the straight line groups intersecting at a predetermined point are the same feature points detected at different times, and the movement of the feature points is detected.

  With reference to FIG. 1 again, the configuration of the feature point movement detection device 1 of the present embodiment will be described. The straight line calculation unit 18 reads an offset value corresponding to the time when the feature point is detected from the inclination table 20, and uses the read offset value to obtain a straight line passing through the detected feature point.

  FIG. 5 is a diagram illustrating an example of data stored in the tilt table 20. FIG. 6 is a diagram for explaining data stored in the tilt table 20. As shown in FIG. 6, the tilt table 20 stores an offset value for the y value.

  As shown in FIG. 5, the inclination table 20 stores offset values for y values for forming five straight lines having different inclinations. For example, a straight line having an inclination θ1 is formed by the offset value of the column with P = 1, and a straight line having the inclination θ2 is formed by the offset value of the column having P = 2. By reading the offset value stored in the inclination table 20 and adding it to the coordinate (x, 0) of the feature point, the coordinate of each point constituting the straight line can be obtained.

  FIG. 7 is a diagram showing the configuration of the memory 24. The memory 24 has a counter corresponding to each point in the coordinate system. The writing control unit 22 counts up the counter of the coordinates at the position to which the offset value read from the tilt table 20 is added. In FIG. 7, the coordinates corresponding to the points c1, c3, c5, c7, and c9 are indicated by hatching. In addition, you may employ | adopt the structure which allocates one physical address to the counter corresponding to each coordinate shown in FIG.

  For example, in the case of a straight line passing through the point c1, the offset value is read from the column “P = 1” of the inclination table 20 shown in FIG. Specifically, since the offset value is “0” when y = 1, the counter of the coordinate immediately above the point c1 is counted up to “1”. In FIG. 7, a counter having a value of 1 or more is shown surrounded by a thick frame. Since the offset value is “1” when y = 2, the counter of the coordinate shifted by one to the right from the point c1 is counted up. Similarly, the counter corresponding to the obtained coordinates is counted up by adding the offset value. In the case of a straight line passing through the point c3, the offset value is read from the column “P = 2” of the inclination table 20 shown in FIG. 5, and the counter corresponding to the coordinate is counted up. Similarly, the detected points c1, c3, c5, c7, and c9 are counted up by adding an offset value determined according to the detection timing. As a result, as shown in FIG. 7, values are input to the counters corresponding to the respective coordinates. As described above, the counter 24 counts up in accordance with the offset value read from the tilt table 20, whereby linear data is stored in the memory 24.

  The memory 24 includes a first area 241, a second area 242,... A fifth area 245 (hereinafter, these areas are collectively referred to as “areas”). Each of the areas 241 to 245 has a counter corresponding to the coordinates as shown in FIG. In each area, data of straight lines whose detection is started at different times are written.

  FIG. 8 is a diagram showing an example of the timing of writing a straight line to each region. As shown in FIG. 8, a straight line passing through the points c1 and c2 detected at time t1 is written in the first region 241. Writing to the first area 241 is started at time t1. At this time, data is not written in the second area 242 and the third area 243.

  Next, the points c3 and c4 detected at time t2 are written in the first area 241 and the second area 242, and not written in the third area 243. Writing to the second area 242 is started at time t2. At this time, in the first region 241, the straight line passing through the points c3 and c4 is a straight line having a different slope from the straight line passing through the already written points c1 and c2. Specifically, for the points c1 and c2, the offset value read from the P = 1 column of the inclination table 20 is used, and for the points c3 and c4, the P = 2 column of the inclination table 20 is used. The counter of the first area 241 is counted up using the read offset value. A straight line having the same inclination as the straight line passing through the points c1 and c2 in the first region 241 is written in the straight line passing through the points c3 and c4 in the second region 242. That is, the counter of the second area 242 is counted up using the offset value read from the column of P = 1 in the inclination table 20.

  Although not shown, the points c5 and c6 are written in the third region 243 at time t3. Writing to the third area 243 is started at time t3.

  At time t5, points c9 and c10 are written in the first area 241, the second area 242, and the third area 243, respectively. Although not shown in FIG. 8, points c9 and c10 are also written in the fourth area 244 and the fifth area 245. At this time, since the straight line with respect to the point detected in five detection timings of the time t1-t5 was written in the 1st area | region 241, correspondence of the points c1-c10 is calculated | required from the intersection state of this straight line group. In this example, it can be seen that the points c1, c3, c5, c7, and c9 are the same feature points, and the points c2, c4, c6, c8, and c10 are the same feature points.

  At time t6, in the first area 241, the data up to time t5 is erased and new points c11 and c12 are written. In the second area 242, the data of the points c11 and c12 are written. At this time, since the straight line with respect to the point detected in five detection timings of the time t2-t6 was written in the 2nd area | region 242, the response | compatibility of the points c3-c12 is calculated | required from the intersection state of this straight line group.

  At time t7, in the second area 242, the data up to time t6 is erased and new points c13 and c14 are written. In the third area 243, data of points c13 and c14 are written. At this time, since the straight line with respect to the point detected in five detection timings of the time t3-t7 was written in the 3rd area | region 243, the response | compatibility of the points c5-c14 is calculated | required from the intersection state of this straight line group.

  Thus, by writing the feature points that have been detected from different timings into the plurality of areas of the memory 24, the movement of the feature points can be detected at different timings in each area. When only one area is used, after detecting the movement of the feature point from the data detected from time t1 to time t5, the next result cannot be obtained unless waiting for the same time as time t1 to time t5. By preparing a plurality of areas 241 to 245 in the memory 24 as in the present embodiment and using the plurality of areas, it is possible to detect the movement of feature points at time t6, t7, etc. following time t5. In the present embodiment, the correspondence is obtained from the feature points detected at the five detection timings, but since the same number of five regions as the number of detections necessary for detecting the movement of the feature points are used, the first region is used. The movement of the feature point can be detected in any one of the areas 241 to 245.

  Further, after detecting the movement of the feature point, the data in the area is cleared, so that the counter value corresponding to each coordinate does not overflow.

  The movement detection unit 26 detects the movement of the feature point based on the intersection state of the straight line group from the data written in each memory 24. The movement detection unit 26 determines whether or not the straight lines intersect by obtaining a location where the intersecting points are concentrated.

  FIG. 9A and FIG. 9B are diagrams for explaining the detection of a concentrated point by the movement detection unit 26. FIG. 9A is a diagram illustrating an example of data written in the memory 24 by the write control unit 22. The movement detection unit 26 obtains the frequency distribution of the counter value from the histogram and determines the crossing state of the straight lines. Here, an example of obtaining a histogram of a portion surrounded by a frame shown in FIG. 9A will be described.

  FIG. 9B is a diagram illustrating an example in which the histogram h is obtained by rotating the portion surrounded by the frame f in FIG. 9A by 90 °. The coordinates corresponding to the points constituting each straight line shown in FIG. 9B have counter values (see FIG. 7). A histogram h is obtained by summing up the values of the counters. The portion where the value of the histogram h is maximum is determined as the point where the intersections of the straight line group are concentrated.

  The result display unit 28 displays the detection result of the feature point movement obtained by the movement detection unit 26. At this time, since the distance between the feature point and the intersection represents the moving speed of the feature point, the result display unit 28 may display the distance between the feature point and the intersection as the moving speed of the feature point. Thereby, the result of detecting the intersection can be used as the speed of the feature point as it is, so that the moving speed of the feature point can be easily obtained.

  Next, the operation of the feature point movement detection device 1 of the present embodiment will be described. FIG. 10 is a diagram illustrating the operation of the feature point movement detection device 1 according to the present embodiment. The feature point movement detection apparatus 1 receives an input of an image at the image input unit 10 (S10). The image input unit 10 inputs the input image to the edge image generation unit 12. The edge image generation unit 12 generates a vertical edge image from the input image (S12), and inputs the generated vertical edge image to the feature point detection unit. The feature point detection unit 14 reads out pixel data in the horizontal direction from the input vertical edge image, and extracts feature points from the read out pixel data (S14). The feature point detection unit 14 inputs the vertical edge image from which the feature point has been detected to the feature point coordinate calculation unit 16. The feature point coordinate calculation unit 16 sets a predetermined coordinate system for the vertical edge image, and calculates the coordinates of the feature point in the predetermined coordinate system (S16). The feature point coordinate calculation unit 16 inputs the calculated feature point coordinates to the straight line calculation unit 18.

  The straight line calculation unit 18 determines the slope of the straight line (S18). The slope of the straight line is determined for each area of the memory 24 to which data is written based on the timing at which the feature points are detected. In FIG. 8, data at time t5 will be described as an example. In the area 241, points c1 and c2 detected at time t1, points c3 and c4 detected at time t2, points c5 and c6 detected at time t3, points c7 and c8 detected at time t4, and time Points c9 and c10 detected at t5 are described. It can be seen that the straight lines passing through the respective points c1 to c10 have an inclination determined according to the detected timing. Also, referring to the second region 242, it can be seen that the straight line passing through the points c3 to c10 has an inclination determined according to the detected timing.

  The straight line calculation unit 18 reads the offset value from the tilt table 20 according to the determined tilt, and obtains the coordinates of each point constituting the straight line (S20). The straight line calculation unit 18 inputs the coordinate value of each point constituting the straight line to the writing control unit 22. Thereafter, the writing control unit 22 counts up the counter value corresponding to each coordinate input from the straight line calculation unit 18, and writes a straight line in each of the areas 241 to 245 (S22).

The movement detection unit 26 selects an area in which feature points for five detection times are written from the areas 241 to 245 of the memory 24. The movement detection unit 26 obtains correspondence between a plurality of feature points based on the intersection state of the straight lines written in the selected area, and detects the movement of the feature points (S24). The writing control unit 22 erases the data in the area used for the movement detection of the feature points by the movement detection unit 26 (S26). The result display unit 28 displays the movement of the feature point detected by the movement detection unit 26 (S28).
The feature point movement detection device 1 and the feature point movement detection method of the present embodiment have been described above.

  The feature point movement detection apparatus 1 according to the present embodiment draws a straight line having an inclination determined based on the detected timing for a plurality of feature points detected at different times, and based on the intersection state of the straight lines. To find the correspondence of the feature points. Thereby, even when a plurality of different feature points are included at the same timing, the movement of the feature points can be detected without performing a complicated calculation process.

  In addition, in the feature point movement detection device 1 according to the present embodiment, the memory 24 includes the first region 241 to the fifth region 245, and stores data of feature points that have been detected from different times in the respective regions. Thereby, in the first region 241 to the fifth region 245, the feature point detection results for five times necessary for obtaining the feature points are sequentially accumulated, so that the movement of the feature points is detected in any region. be able to.

  In addition, the feature point movement detection device 1 according to the present embodiment stores straight line data in a form in which a counter corresponding to coordinates is counted up, and obtains a frequency distribution of the counter values, whereby the intersections of straight lines are concentrated. Therefore, it is not necessary to calculate the crossing point from the straight line expression, and the straight line crossing state can be easily obtained.

  Further, after detecting the movement of the feature point, the feature point movement detection device 1 according to the present embodiment erases the data in the area storing the data used to detect the movement of the feature point. Can be prevented from overflowing.

  The feature point movement detection device 1 of the present invention has been described in detail with reference to the embodiment, but the feature point movement detection device 1 of the present invention is not limited to the above-described embodiment.

  In the above-described embodiment, the straight line data is stored in the memory 24 for storing the straight line data by preparing a counter corresponding to the coordinates and storing the counter value. Also good. In this case, the capacity of the memory 24 for storing linear data can be reduced.

  In the above-described embodiment, the feature points are detected at five detection timings, and the correspondence of the feature points obtained at different detection timings is obtained by drawing five types of straight lines on the detected feature points. The number of detections necessary for obtaining the correspondence between feature points is not limited to five. For example, the movement of the feature point may be detected by drawing seven types of straight lines on the feature point detected at seven detection timings. As the number of detections increases, the accuracy with which the movement of feature points can be detected increases. If the movement of the feature point is detected with a small number of detections, the movement of the feature point can be detected quickly with a small amount of data. It should be noted that it is preferable to prepare the same number of areas in the memory as the number of detection times used for feature point movement detection. Thereby, when writing is started in each area at different timings, it is possible to detect the movement of the feature point in each area sequentially.

  In the above-described embodiment, the feature point detection timing interval is constant, and the straight line is calculated based on the offset value read from the inclination table 20, but the present invention is also applied to the case where the feature point detection timing interval is not constant. Can be applied. In this case, the inclination is determined based on the detection timing interval. For example, when the detection timing interval changes from halfway to the middle, the slope θ of the straight line is set so that the change in tanθ is doubled.

  In the above-described embodiment, the example of detecting the movement of the feature point moving in the horizontal direction has been described, but the movement direction of the feature point that can be detected is not limited to the horizontal direction, and the movement in the vertical direction can also be detected. It is also possible to determine the movement of the feature point in an arbitrary direction based on the movement of the feature point in the horizontal direction and the vertical direction.

  In the above-described embodiment, the feature point movement detection device 1 and the operation thereof have been described. However, the present invention includes a program for realizing the function of the feature point movement detection device 1 described above.

  According to the present invention, it is possible to detect the movement of a feature point by a simple process of obtaining a straight line of an inclination determined based on the detection timing for a feature point detected from data. For example, the present invention is useful as a moving object detection device or the like that acquires in real time motion information in video captured by a high-speed camera or the like. Also, for example, in a car or the like, it is also useful for detecting the jumping out of other cars or children.

It is a figure which shows the structure of the feature point movement detection apparatus of this Embodiment. It is a figure which shows the example of a vertical edge image. It is a figure which shows a mode that a feature point moves. It is a figure which shows the principle which calculates | requires the response | compatibility of the feature point detected at different time. It is a figure which shows the example of the data memorize | stored in the inclination table. It is a figure for demonstrating the data memorize | stored in the inclination table. It is a figure which shows the structure of a memory. It is a figure which shows the example of the write-in timing of the straight line to each area | region. (A) It is a figure which shows the example of the data written in memory by the write control part. (B) It is a figure which shows the example which calculated | required the histogram of the counter value. It is a figure which shows operation | movement of the feature point movement detection apparatus of this Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Feature point detection apparatus 10 Image input part 12 Edge image generation part 14 Feature point detection part 16 Feature point coordinate calculation part 18 Straight line calculation part 20 Inclination table 22 Write control part 24 Memory 26 Movement detection part 28 Result display part

Claims (7)

An apparatus for detecting movement of a feature point,
A feature point detector that detects feature points from the input data;
A straight line calculation unit that obtains a straight line that passes through the feature point detected by the feature point detection unit and has an inclination determined based on the timing at which the feature point is detected;
For feature points detected at different timings, a memory for storing a straight line calculated by the straight line calculation unit;
A writing control unit for writing the straight line calculated by the straight line calculating unit into the memory;
Based on the intersection state of the plurality of straight lines stored in the memory, a movement detection unit that obtains correspondence between the plurality of feature points detected at different timings and detects movement of the feature points;
A feature point movement detection device comprising: The memory prepares a counter corresponding to each coordinate of a predetermined coordinate system that defines the straight line,
The writing control unit counts up a counter corresponding to the coordinates of each point constituting the straight line calculated by the straight line calculating unit,
The movement detection unit obtains a crossing state of straight lines based on a counter value corresponding to each coordinate of the memory;
The feature point movement detection apparatus according to claim 1. The memory has a plurality of areas,
The write control unit starts writing data from each of a plurality of areas at different timings,
3. The feature point movement detection device according to claim 1, wherein the movement detection unit detects movement of the feature point with reference to an area in which the number of data writing reaches a predetermined number among a plurality of areas.   The feature point movement detection device according to claim 3, wherein the writing control unit erases data in a region used by the movement detection unit to detect movement of the feature point.   The feature point movement detection device according to claim 1, wherein the straight line calculation unit sets the slope θ of the straight line so that a change in tan θ is proportional to an interval of detection timing of feature points. A method for detecting movement of a feature point,
A feature point detection step of detecting feature points from the input data;
A straight line calculating step for obtaining a straight line that passes through the characteristic point detected in the characteristic point detecting step and has an inclination determined based on the timing of detecting the characteristic point;
For the feature points detected at different timings, preparing a memory for storing the straight line calculated in the straight line calculating step;
A writing control step of writing the straight line calculated in the straight line calculating step into the memory;
A movement detection step for obtaining correspondences of the plurality of feature points detected at different timings based on the intersection states of the plurality of straight lines stored in the memory, and detecting movement of the feature points;
A feature point movement detection method comprising: A program for detecting the movement of a feature point,
A feature point detection step of detecting feature points from the input data;
A straight line calculating step for obtaining a straight line that passes through the characteristic point detected in the characteristic point detecting step and has an inclination determined based on the timing of detecting the characteristic point;
Generating a memory area for storing a straight line calculated in the straight line calculating step for feature points detected at different timings;
A writing control step of writing the straight line calculated in the straight line calculating step into the memory area;
A movement detection step for obtaining correspondence between the plurality of feature points detected at different timings based on the intersection state of the plurality of straight lines stored in the memory area, and detecting movement of the feature points;
A program that executes

Images