JP3482717B2 - Image target detection device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to image target detection for automatically detecting a moving object from an image having a complicated background obtained when an image of a vehicle traveling on the ground or a flying object flying at low altitude is imaged from above. It relates to the device.

[0002]

2. Description of the Related Art When a vehicle or building traveling on the ground from a distance is imaged, it looks like small dots whose shape is not clear on the screen, and it is difficult to identify the target object from the shape of the image. In order to detect the target object from such a screen, it is considered effective to detect some motion information from the image.

FIG. 11 is a block diagram of a conventional image target detecting apparatus. In FIG. 11, 1 is an image pickup device, 2 is a point image enhancement filter, 3 is a position measuring unit, 4 is a position storing unit, and 5 is an associating unit. , 6 are moving target detectors.

The image signal from the image pickup device 1 enhances a point image which is a target candidate in the point image enhancement filter 2. This process suppresses the gradual brightness gradient of the background,
This is done to facilitate later position measurement.
As a typical filter for emphasizing the point image, a Laplacian filter showing the effect of the second derivative can be cited, but any filter other than the Laplacian filter may be used as long as it has the effect of emphasizing the point image. For the image in which the point image is emphasized, the position measuring unit 3 measures the position of the point image. Regarding the position of the point image, the image with the point image emphasized is thresholded with an appropriate threshold value to obtain a binary image, and the position of the center of gravity of the connected region of the binary image is set as the position of the point image. Can be measured at. In addition, here, a block formed by significant pixels of the binary image is referred to as a connected region. The position of the point image measured in this way is stored in the position storage unit 4. The same process is performed on the screens that the image pickup device 1 outputs one after another. When looking at the ground from a distant sky, there are many point images of trees and buildings. The associating unit 5 associates many point images existing on each screen with the same point image on the next screen. The method of association is shown in FIG. In FIG. 9, 7 is a screen and 8
a, 8b and 8c are the positions of the point image on a screen, 9a,
9b and 9c represent the positions of point images on the next screen. As shown in FIG. 9, the point images on the next screen within a predetermined range r from the position of the point image are 8a, 9a, 8
b and 9b are associated with each other, and 8c and 9c are associated with each other. The moving target detection unit 6 observes the corresponding point images for a certain period of time, checks each moving amount, and determines a point image in which the moving amount exceeds a predetermined determination value as a moving target.

[0005]

When the image pickup device 1 is moving at high speed, an object other than a moving object that is fixed to the ground may not appear to be stationary in one place on the screen. . FIG. 10 is a diagram for explaining a problem of the conventional technique, and 10 is a screen showing a state in which the ground is imaged from obliquely above the sky, and the ground is represented by a grid. Considering the case where the image pickup device 1 is approaching the ground, the field of view after the time when the screen 10 is imaged is reduced by the same magnification in the vertical and horizontal directions, and the ground reflected on the screen is enlarged. Looks like 12
Is an object fixed to the ground, and 13 is a moving body moving with respect to the ground. Since both the object 12 and the moving body 13 move on the screen with time, the conventional image target detecting device erroneously detects a fixed object on the ground as a moving target. In addition, even if some vibration is applied to the image pickup device 1 and the image is blurred, a fixed object is erroneously detected as a moving target.

The present invention has been made in order to solve such a problem, and by moving the image pickup device,
It is an object of the present invention to provide an image target detecting device capable of detecting only a moving target in an image even if the background appears to be translated, enlarged or reduced.

[0007]

Embodiment 1 of the present invention
The image target detecting apparatus according to 3, 4, 5 and 6 includes an image pickup device which outputs an image signal, a point image enhancement filter which receives an image signal output from the image pickup device and emphasizes a point image included in the image signal, and an image pickup device. Point image removal filter that removes the point image contained in the image signal by inputting the image signal output by the detector
A position measurement unit that finds the position of the point image on the screen based on the output of the point image enhancement filter, a position storage unit that stores the position of the point image, and a point image detected on one screen are detected on the next screen. An associating unit for associating with the point image , and the point image
A first image storage unit that stores an image from which the point image from the removal filter has been removed, a second image storage unit that stores an image from which the point image has been removed from the point image removal filter , and a screen divided into rectangles. A rectangular area whose position on the screen is the same as that of another rectangular area including the rectangular area read by the first read controller and the first read controller that reads the rectangular area from the first image storage unit. An image correlation for determining a location where a rectangular area output by the first read control unit and a second read control unit that reads out from the second image storage unit match in the rectangular area output by the second read control unit Arithmetic unit, effective displacement determination unit that determines effective data from displacements of two screens output from image correlation arithmetic unit, smoothing processing unit that smoothes effective displacement, position of point image and output of smoothing processing unit The displacement of the image Only provided a moving target detection unit for detecting a moving target, said first image storage unit, once the serial image
After memorizing, the moving target detection unit judges the moving target.
Continue to hold the stored image until
The image storage section stores a new image each time the image on the next screen is input.
The stored contents are updated with the image of the surface .

The image target detecting apparatus according to the second embodiment of the present invention includes an image pickup device that outputs an image signal, and a point image removal device that receives an image signal output from the image pickup device and removes a point image contained in the image signal. A filter, a subtraction unit that subtracts the point image removed image from the image signal output from the imager, a position measurement unit that determines the position of the point image on the screen based on the output of the subtraction unit, and a position that stores the position of the point image A storage unit, an associating unit that associates a point image detected on one screen with a point image detected on the next screen, and an image from which the point image is removed from the point image removal filter is stored. A first image storage section,
A second image storage unit that stores an image from which the point image is removed from the point image removal filter; a first read control unit that reads a rectangular area of a screen divided into rectangles from the first image storage unit; A second read control unit for reading a rectangular area whose position on the screen is the same as another rectangular area including the rectangular area read by the first read control unit from the second image storage unit; Effective data is determined from the image correlation calculation unit that determines where the rectangular area output by the control unit matches within the rectangular region output by the second read control unit, and the displacement of the two screens output by the image correlation calculation unit the effective displacement determination unit that, a smoothing processing section for smoothing the effective displacement and a moving target detection unit that detects a moving target by receiving the displacement of the output image of the position and the smoothing processing unit of the point image is provided, before
The first image storage unit stores the image once and then
It is described until the moving target detection section finishes determining the moving target.
The second image storage unit continues to hold the stored image,
Each time a screen image is input, a new screen image is stored
The contents are updated .

[0009]

In the image target detecting apparatus according to the first, third and fourth embodiments of the present invention, the point image included in the image signal is emphasized, the position of the point image on the screen is obtained, and the point image detected on a certain screen is detected. At the same time as the point image detected on the next screen, the point image including the moving object image is removed, the displacement of two screens is calculated for each rectangular area of the screen divided into rectangles, and the effective displacement is smoothed. Then, the movement target is obtained, and the movement target is determined from the movement amount of the point image with respect to the background.

In the image target detecting apparatus according to the second embodiment of the present invention, the point image included in the image signal is emphasized by using the image from which the point image is removed, and the position of the point image on the screen is obtained. At the same time as matching the point image detected on one screen with the point image detected on the next screen, the point image containing the moving object image is removed, and the displacement of two screens for each rectangular area of the screen divided into rectangles. The movement is calculated, the effective displacement is smoothed to obtain the background movement, and the movement target is determined from the movement amount of the point image with respect to the background.

Further, in the image target detecting apparatus according to the fifth embodiment of the present invention, the point image included in the image signal is emphasized, the position of the point image on the screen is determined from the position of the center of gravity of the connected region, and the connected region is detected. Limit the number of point image data based on the brightness, and make the point image detected on one screen correspond to the point image detected on the next screen, and at the same time, remove the point image including the moving object image to make a rectangle. The displacement of the two screens is calculated for each of the divided rectangular areas of the screen, the effective displacement is smoothed to obtain the movement of the background, and the movement target is determined from the movement amount of the point image with respect to the background.

Further, in the image target detecting apparatus according to the sixth embodiment of the present invention, the point image included in the image signal is emphasized, and the position of the point image on the screen is obtained from the position of the maximum point,
The number of point image data is limited based on the brightness of the maximum point, and the point image detected on one screen is made to correspond to the point image detected on the next screen, and at the same time, the point image including the moving image is removed. , The displacement of the two screens is calculated for each rectangular area of the screen divided into rectangles, the effective displacement is smoothed to obtain the movement of the background, and the movement target is determined from the movement amount of the point image with respect to the background.

[0013]

【Example】

Example 1. 1 is a block diagram of an image target detecting apparatus according to a first embodiment of the present invention. In FIG. 1, 1 is an image pickup device, 2 is a point image enhancement filter, 3 is a position measurement unit, 4 is a position storage unit, 5 is an association unit, 14 is a point image removal filter, and 15
Is a first image storage unit, 16 is a second image storage unit, 17 is a first read control unit, 18 is a second read control unit,
Reference numeral 19 is an image correlation calculation unit, 20 is an effective displacement determination unit, 21 is a smoothing processing unit, and 22 is a moving target detection unit.

The image signal from the image pickup device 1 enhances the point image which is a target candidate in the point image enhancement filter 2. This is done in order to suppress the gradual inclination of the brightness of the background and facilitate the subsequent position measurement. As a typical filter for emphasizing the point image, a Laplacian filter showing the effect of the second derivative can be cited, but any filter other than the Laplacian filter may be used as long as it has the effect of emphasizing the point image. For the image in which the point image is emphasized, the position measuring unit 3 measures the position of the point image. The measured position of the point image is stored in the position storage unit 4.

Similar processing is performed on the screens output by the image pickup device 1 one after another. When looking at the ground from a distant sky, there are many point images of trees and buildings. Association unit 5
Makes many point images existing on each screen correspond to the same point image on the next screen. The associating method is the same as the method shown in the embodiment of the conventional image target detecting apparatus, and as shown in FIG. 9, the point image on the next screen within a predetermined range r from the position of the point image is the same object. 8a and 9a, 8b
And 9b and 8c and 9c are made to correspond.

The point image removal filter 14 receives the image signal of the image pickup device 1 and removes the point image. This is a process necessary to prevent an error from occurring in the detection of the background movement obtained by the image correlation calculation unit 19 under the influence of the moving image included in the image. As a point image removal filter,
Although a median filter can be used, a filter other than the median filter can be used as long as it has a similar effect. The image signal from which the point image has been removed is stored in the first image storage unit 15 and the second image storage unit 16. After first storing the image, the first image storage unit 15 continues to hold the stored image until the determination of the moving target detection unit 22 described later is completed, and the second image storage unit 16 stores the next image. Every time the screen image of is input, the stored contents are updated with the new screen image.

Next, the first read control unit 17, the second read control unit 18, and the image correlation calculation unit 19 obtain the movements of the background images at a plurality of places on the screen by using the correlation calculation. FIG. 7 is a diagram showing a section on the screen which is a target of the correlation calculation, and shows a state in which it is divided into four in the vertical direction and five in the horizontal direction, but the number of divisions may be other than that.
In FIG. 7, reference numeral 23 is a reference image for correlation calculation, and 24 is a search area for scanning the reference image to search for a matching position.
The first read control unit 17 reads the reference image 23 from the first image storage unit 15 and outputs it to the image correlation calculation unit 19. The second read control unit 18 uses the second image storage unit 1
The search area 24 is read from 6 and output to the image correlation calculation section 19. Each read control unit outputs the reference image and the search image of another section to the image correlation calculation unit 19 after waiting for the calculation of the image correlation calculation unit 19 to be ready to receive the next data. The image correlation calculation unit 19
The position of the search area that can be matched with the reference image for each section for image correlation is output as the displacement from the position of the reference image.

Now, FIG. 8 shows a diagram in which the displacement of each section output by the image correlation calculation unit 19 is represented by a vector. When the image pickup device 1 is close to the ground, the image expands with time and the displacement vector spreads outward as shown in FIG. 8. However, in a situation where there is noise in the image, the correct displacement like the vector 25 is required. It's possible that nothing happens. The effective displacement determination unit 20 is a processing unit for detecting such an erroneous displacement. As a method of determining an abnormal displacement, for example, a displacement whose direction is significantly different from that of a surrounding section is determined to be invalid.

The smoothing processing unit 21 smoothes the background movement amount by calculating the average value of the effective displacements in the vicinity output by the effective displacement determination unit 20 for each section.

The moving target detecting unit 22 detects the moving target from the moving amount of the target candidate sent from the associating unit 5 and the background moving amount sent from the smoothing processing unit 21. As the amount of movement of the background, the amount of movement of the section according to the position of the target candidate is used. The determination of the moving target is performed by determining the moving target when the magnitude of the difference vector between the background moving vector and the moving vector of the target candidate exceeds the determination value in consideration of each measurement error.

Example 2. 2 is a block diagram of an image target detecting apparatus according to a second embodiment of the present invention. In FIG. 2, 1 is an image pickup device, 2 is a point image enhancement filter, 3 is a position measuring unit, 4
Is a position storage unit, 5 is an association unit, 26 is a subtraction unit, 14 is a point image removal filter, 15 is a first image storage unit, 16 is a second image storage unit, 17 is a first read control unit, 18
Is a second read control unit, 19 is an image correlation calculation unit, 20
Is an effective displacement determination unit, 21 is a smoothing processing unit, and 22 is a moving target detection unit.

The point image is emphasized by using the subtractor 26 and the point image removing filter 14. The subtractor 26 subtracts, for each pixel, the image from which the point image has been removed by the point image removal filter 14 from the image signal of the image pickup device 1. As a result, an image in which the point image is emphasized is obtained by suppressing the gradual luminance gradient of the background. The position measuring unit 3, the first image storage unit 15 and the second image storage unit 16 and thereafter, the moving target detection unit 2
The operations up to 2 are exactly the same as in the first embodiment.

Example 3. The overall configuration of the image target detecting apparatus according to the third embodiment of the present invention is the same as that of the first embodiment.

The image signal from the image pickup device 1 enhances a point image which is a target candidate in the point image enhancement filter 2. This is done in order to suppress the gradual inclination of the brightness of the background and facilitate the subsequent position measurement. As a typical filter for emphasizing the point image, a Laplacian filter showing the effect of the second derivative can be cited, but any filter other than the Laplacian filter may be used as long as it has the effect of emphasizing the point image. For the image in which the point image is emphasized, the position measuring unit 3 measures the position of the point image. The measured position of the point image is stored in the position storage unit 4.

Similar processing is performed on the screens output by the image pickup device 1 one after another. When looking at the ground from a distant sky, there are many point images of trees and buildings. Association unit 5
Makes many point images existing on each screen correspond to the same point image on the next screen. The associating method is the same as the method shown in the embodiment of the conventional image target detecting apparatus, and as shown in FIG. 9, the point image on the next screen within a predetermined range r from the position of the point image is the same object. 8a and 9a, 8b
And 9b and 8c and 9c are made to correspond.

The point image removing filter 14 receives the image signal of the image pickup device 1 and removes the point image. This is a process necessary to prevent an error from occurring in the detection of the background movement obtained by the image correlation calculation unit 19 under the influence of the moving image included in the image. As a point image removal filter,
Although a median filter can be used, a filter other than the median filter can be used as long as it has a similar effect. The image signal from which the point image has been removed is stored in the first image storage unit 15 and the second image storage unit 16. After first storing the image, the first image storage unit 15 continues to hold the stored image until the determination of the moving target detection unit 22 described later is completed, and the second image storage unit 16 stores the next image. Every time the screen image of is input, the stored contents are updated with the new screen image.

Next, the first read control unit 17, the second read control unit 18, and the image correlation calculation unit 19 obtain the movements of the background images at a plurality of places on the screen by using the correlation calculation. FIG. 7 is a diagram showing a section on the screen which is a target of the correlation calculation, and shows a state in which it is divided into four in the vertical direction and five in the horizontal direction, but the number of divisions may be other than that.
In FIG. 7, reference numeral 23 is a reference image for correlation calculation, and 24 is a search area for scanning the reference image to search for a matching position.
The first read control unit 17 reads the reference image 23 from the first image storage unit 15 and outputs it to the image correlation calculation unit 19. The second read control unit 18 uses the second image storage unit 1
The search area 24 is read from 6 and output to the image correlation calculation section 19. Each read control unit outputs the reference image and the search image of another section to the image correlation calculation unit 19 after waiting for the calculation of the image correlation calculation unit 19 to be ready to receive the next data. The image correlation calculation unit 19
The position of the search area that can be matched with the reference image for each section for image correlation is output as the displacement from the position of the reference image.

FIG. 8 shows a vector representation of the displacement of each section output by the image correlation calculation unit 19. When the image pickup device 1 is close to the ground, the image expands with time and the displacement vector spreads outward as shown in FIG. 8. However, in a situation where there is noise in the image, the correct displacement like the vector 25 is required. It's possible that nothing happens. The effective displacement determination unit 20 is a processing unit for detecting such an erroneous displacement.

FIG. 3 is a block diagram of an effective displacement judging section of an image target detecting apparatus according to a third embodiment of the present invention. In FIG. 3, 27 is a median value calculation unit, 28 is a storage unit, 29 is a subtraction unit,
30 is a determination unit.

Whether the displacement is valid or not is determined separately for the x component and the y component. To determine the validity of the x component of the displacement, the x component of the displacement of the group 31 of vertically continuous sections is used as shown in FIG. The median value calculation unit 27 obtains the median value of the x component of the group of sections 31. The median refers to the value located at the center when the numerical sequences are arranged in order of size. When the number of sequences is an even number, it is 2 which is the closest to the value obtained by dividing the number of sequences by 2.
It is the value of the rank of either one of the two integers. Storage unit 28
Is a means for temporarily storing the displacement of the cluster 31 of partitions, and the subtraction unit 29 obtains the difference between the displacement of each partition and the median value. The determination unit 30 determines that the x component of the displacement within which the difference obtained by the subtraction unit 29 is within the allowable error range is valid,
The x component of the displacement that exceeds the allowable error range is determined to be invalid. The same determination is made for the second to fifth columns. To determine the effectiveness of the y component of the displacement, the y component of the displacement of the group 32 of laterally continuous sections is used as shown in FIG. The effective displacement determination unit 20 obtains the median value of the y components of the group of sections 32, determines that the y component of the displacement within the allowable error range from the median value is effective, and determines the displacement exceeding the allowable error range. It is determined that the y component is invalid. Similar determination is made for other rows. In the above-described embodiment, the group 31 of vertical sections is used for the determination of the x component, and the group 32 of horizontal sections is used for the determination of the y component. It is not necessary to use it, and the median value may be obtained for the x component and the y component in the group 33 of common sections in FIG.

The smoothing unit 21 smoothes the x component and the y component of the displacement separately. For smoothing the x-component, the average of the effective x-components of the vertical partition set 31 is calculated. This average value is output to the moving target detection unit as the x component of the displacement in the first column. An average is similarly calculated for the second to fifth columns. For the y component of the displacement, calculate the average of the valid y components of the lateral partition collection 32. This average value is output to the moving target detection unit as the y component of the displacement of the second row. The average is similarly calculated for the other rows. When the image is enlarged or reduced, it is effective to average the x components in the vertical columns and the y components in the horizontal rows as in this embodiment. However, it is not always necessary to perform averaging in vertically long or horizontally long columns, and averaging may be performed in the region 33 near that section. By thus determining the effective displacement and calculating the average thereof, it is possible to accurately obtain the movement of the background image between the screens.

The moving target detecting unit 22 detects the moving target from the moving amount of the target candidate sent from the associating unit 5 and the background moving amount sent from the smoothing processing unit 21. As the amount of movement of the background, the amount of movement of the section according to the position of the target candidate is used. The determination of the moving target is performed by determining the moving target when the magnitude of the difference vector between the background moving vector and the moving vector of the target candidate exceeds the determination value in consideration of each measurement error.

Example 4. The overall configuration of the image target detecting apparatus according to the fourth embodiment of the present invention is the same as that of the first embodiment.

The image signal from the image pickup device 1 enhances the point image which is a target candidate in the point image enhancement filter 2. This is done in order to suppress the gradual inclination of the brightness of the background and facilitate the subsequent position measurement. As a typical filter for emphasizing the point image, a Laplacian filter showing the effect of the second derivative can be cited, but any filter other than the Laplacian filter may be used as long as it has the effect of emphasizing the point image. For the image in which the point image is emphasized, the position measuring unit 3 measures the position of the point image. The measured position of the point image is stored in the position storage unit 4.

The same processing is performed on the screens output by the image pickup device 1 one after another. When looking at the ground from a distant sky, there are many point images of trees and buildings. Association unit 5
Makes many point images existing on each screen correspond to the same point image on the next screen. The associating method is the same as the method shown in the embodiment of the conventional image target detecting apparatus, and as shown in FIG. 9, the point image on the next screen within a predetermined range r from the position of the point image is the same object. 8a and 9a, 8b
And 9b and 8c and 9c are made to correspond.

The point image removal filter 14 receives the image signal from the image pickup device 1 and removes the point image. This is a process necessary to prevent an error from occurring in the detection of the background movement obtained by the image correlation calculation unit 19 under the influence of the moving image included in the image. As a point image removal filter,
Although a median filter can be used, a filter other than the median filter can be used as long as it has a similar effect. The image signal from which the point image has been removed is stored in the first image storage unit 15 and the second image storage unit 16. After first storing the image, the first image storage unit 15 continues to hold the stored image until the determination of the moving target detection unit 22 is completed, and the second image storage unit 16 stores the image of the next screen. Whenever is input, the stored contents are updated with a new screen image.

Next, the first read control unit 17, the second read control unit 18, and the image correlation calculation unit 19 obtain the movements of the background images at a plurality of places on the screen by using the correlation calculation. FIG. 7 is a diagram showing a section on the screen which is a target of the correlation calculation, and shows a state in which it is divided into four in the vertical direction and five in the horizontal direction, but the number of divisions may be other than that.
In FIG. 7, reference numeral 23 is a reference image for correlation calculation, and 24 is a search area for scanning the reference image to search for a matching position.
The first read control unit 17 reads the reference image 23 from the first image storage unit 15 and outputs it to the image correlation calculation unit 19. The second read control unit 18 uses the second image storage unit 1
The search area 24 is read from 6 and output to the image correlation calculation section 19. Each read control unit outputs the reference image and the search image of another section to the image correlation calculation unit 19 after waiting for the calculation of the image correlation calculation unit 19 to be ready to receive the next data. The image correlation calculation unit 19
The position of the search area that can be matched with the reference image for each section for image correlation is output as the displacement from the position of the reference image.

Now, FIG. 8 shows a diagram in which the displacement of each section output by the image correlation calculation unit 19 is represented by a vector. When the image pickup device 1 is close to the ground, the image expands with time and the displacement vector spreads outward as shown in FIG. 8. However, in a situation where there is noise in the image, the correct displacement like the vector 25 is required. It's possible that nothing happens. The effective displacement determination unit 20 is a processing unit for detecting such an erroneous displacement. As a method of determining an abnormal displacement, for example, a displacement whose direction is significantly different from that of a surrounding section is determined to be invalid.

FIG. 4 is a block diagram of a smoothing processing section of an image target detecting apparatus according to a fourth embodiment of the present invention. 3 in FIG.
4 is a first average value calculation unit, 35 is an effective displacement counting unit, 36
Is a storage unit, 37 is a second average value calculation unit, and 38 is a selection unit.

The smoothing section 21 smoothes the x component and the y component of the displacement separately. For smoothing the x component, the first average value calculator 34 calculates the average of the effective x components of the group 31 of vertical sections. This average value is used as the x-component of the displacement of the first column and is passed through the selection unit 38 to the moving target detection unit
Output to 2. An average is similarly calculated for the second to fifth columns. For the y component of the displacement, the average value calculation unit 34 calculates the average of the effective y components of the group 32 of lateral sections. This average value is output to the moving target detection unit 22 via the selection unit 38 as the y component of the displacement of the second row. The average is similarly calculated for the other rows. When the image is enlarged or reduced, it is effective to average the x components in the vertical columns and the y components in the horizontal rows as in this embodiment. However, it is not always necessary to perform averaging in vertically long or horizontally long columns, and averaging may be performed in the region 33 near that section. The average value of the effective displacements for each column and each row is stored in the storage unit 36. The effective displacement counting unit 35 counts the number of effective displacements for each column and each row. Here, when there are few effective displacements in the set of sections to be averaged, for example, when three of the four displacements in the second row are invalid,
As the displacement of the second column, the average value of the second column, that is, the only valid displacement value itself is not output, and the first value is stored in the storage unit 36.
The average value of the column and the third column is read out, and the second average value calculation unit 3
7, the average values of the first and third columns are further averaged to obtain the output value of the second column. The selection unit 38 selects and outputs the output of the second average value calculation unit 37 instead of the first average value calculation unit 34 when the effective displacement is small in the set of sections to be averaged. By thus determining the effective displacement and calculating the average thereof, it is possible to accurately obtain the movement of the background image between the screens.

The moving target detecting unit 22 detects the moving target from the moving amount of the target candidate sent from the associating unit 5 and the background moving amount sent from the smoothing processing unit 21. As the amount of movement of the background, the amount of movement of the section according to the position of the target candidate is used. The determination of the moving target is performed by determining the moving target when the magnitude of the difference vector between the background moving vector and the moving vector of the target candidate exceeds the determination value in consideration of each measurement error.

Example 5. The overall configuration of the image target detecting apparatus according to the fifth embodiment of the present invention is the same as that of the first embodiment.

The image signal from the image pickup device 1 enhances the point image which is a target candidate in the point image enhancement filter 2. This is done in order to suppress the gradual inclination of the brightness of the background and facilitate the subsequent position measurement. As a typical filter for emphasizing the point image, a Laplacian filter showing the effect of the second derivative can be cited, but any filter other than the Laplacian filter may be used as long as it has the effect of emphasizing the point image. For the image in which the point image is emphasized, the position measuring unit 3 measures the position of the point image.

As for the position of the point image, an image in which the point image is emphasized is thresholded with an appropriate threshold value to obtain a binary image,
It can be measured by setting the position of the center of gravity of the connected region of the binary image as the position of the point image. In addition, here, a block formed by significant pixels of the binary image is referred to as a connected region. Figure 5
6 is a configuration diagram showing an embodiment of the position measuring unit 3, 39 in FIG. 5 is a binarization circuit, 40 is a connected region center of gravity measuring unit, 4
Reference numeral 1 is a connected area luminance measuring unit, and 42 is a sorting unit. The image signal 43 emphasizing the dot image is converted into a binarization circuit 39.
Is converted into a binary image signal 44 by thresholding with an appropriate threshold. The connection area center of gravity measuring unit 40 measures the position of the center of gravity of the connection area. The connected region brightness measuring unit 41 measures the maximum value of the brightness of the pixels included in the connected region for each connected region and outputs the maximum value to the sorting unit 42. The sorting unit 42 receives the barycentric position of each connected region and the maximum brightness, arranges the barycentric positions of the connected regions in descending order of brightness, and outputs a predetermined number of barycentric positions of the connected regions in descending order of brightness. When an infrared image pickup device is used as the image pickup device, the target object looks high in brightness, and therefore the above-mentioned sorting is prioritized as a target candidate. In this way, by limiting the number of data of the position of the center of gravity, it is possible to suppress the processing time of the next association without dropping the true target from the target candidates. The position of the point image measured in this way is stored in the position storage unit 4.

After the associating unit 5 and the point image filter 14,
The operation up to the moving target detection unit 22 is exactly the same as in the first embodiment.

Example 6. The entire configuration of the image target detecting apparatus according to the sixth embodiment of the present invention is the same as that of the first embodiment.

The image signal from the image pickup device 1 enhances the point image which is a target candidate in the point image enhancement filter 2. This is done in order to suppress the gradual inclination of the brightness of the background and facilitate the subsequent position measurement. As a typical filter for emphasizing the point image, a Laplacian filter showing the effect of the second derivative can be cited, but any filter other than the Laplacian filter may be used as long as it has the effect of emphasizing the point image. For the image in which the point image is emphasized, the position measuring unit 3 measures the position of the point image.

FIG. 6 is a block diagram showing an embodiment of the position measuring unit 3. In FIG. 6, 42 is a sorting unit and 47 is a maximum point detecting unit. The image signal 43 in which the point image is emphasized is
The position of the maximum point is measured by the maximum point detection unit 47. The sorting unit 42 receives the positions of the maximum points 48 and the brightness of the maximum points, arranges the positions of the maximum points in descending order of the brightness, and outputs a predetermined number of positions of the maximum points from the one having the highest brightness. When an infrared image pickup device is used as the image pickup device, the target object looks high in brightness, and therefore the above-mentioned sorting is prioritized as a target candidate. In this way, by limiting the number of data at the position of the maximum point, it is possible to suppress the processing time of the next association without dropping the true target from the target candidates. The operation other than the position measuring unit 3 is exactly the same as that of the first embodiment.

[0049]

As described above, according to the image target detecting apparatus according to the first, third and fourth embodiments of the present invention, the point image removing filter for removing the point image included in the image signal and the image without the point image are removed. A first image storage unit for storing the image, a second image storage unit for storing the image from which the point image is removed, and a first read control for reading a rectangular area of the screen divided into rectangles from the first image storage unit. And a second read control unit for reading from the second image storage unit a rectangular area having the same position on the screen as another rectangular area including the rectangular area read by the first read control unit, An image correlation calculation unit that determines a position where the rectangular area output by the first read control unit matches within the rectangular area output by the second read control unit;
An effective displacement determination unit that determines effective data from the displacements of the two screens output by the image correlation calculation unit and a smoothing processing unit that smoothes the effective displacement are provided, and the point image including the moving object image is removed to form a rectangle. The displacement of two screens is calculated for each rectangular area of the divided screen, and the effective displacement is smoothed to obtain the background motion,
Configured to compare the movement of the target candidate and the movement of the background,
After the first image storage unit stores the image once,
Until the moving target detection unit determines the moving target
Continuing to hold the stored image, the second image storage unit,
Every time the image of the next screen is input, it will be memorized as the image of the new screen.
Since the contents are updated , the true moving target can be accurately detected even in a situation where the screen appears to move in parallel, expands or contracts.

Further, according to the image target detecting apparatus in accordance with the second embodiment of the present invention, a subtracting section for subtracting the point image removed image from the image signal output from the image pickup device is provided to give the same effect as the point image enhancing filter. Therefore, it is possible to accurately detect a true moving target in a situation where the screen appears to be translated, enlarged, or reduced with a small hardware configuration.

Further, according to the image target detecting apparatus in the fifth embodiment of the present invention, the position of the point image on the screen is obtained from the position of the center of gravity of the connected area, and the number of data of the point image is calculated based on the brightness of the connected area. Since it is limited, the moving target can be detected in a short processing time.

Further, according to the image target detecting apparatus in the sixth embodiment of the present invention, the position of the point image on the screen is obtained from the position of the maximum point and the number of data of the point image is limited based on the brightness of the maximum point. By doing so, the moving target can be detected in a short processing time.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an image target detecting apparatus according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of an image target detecting apparatus according to a second embodiment of the present invention.

FIG. 3 is a configuration diagram of an effective displacement determination unit of an image target detection device according to a third embodiment of the present invention.

FIG. 4 is a configuration diagram of a smoothing processing unit of an image target detecting apparatus according to Embodiment 4 of the present invention.

FIG. 5 is a configuration diagram of a position measuring unit of an image target detecting apparatus according to a fifth embodiment of the present invention.

FIG. 6 is a configuration diagram of a position measuring unit of an image target detecting apparatus according to a sixth embodiment of the present invention.

FIG. 7 is an explanatory diagram showing sections for performing image correlation.

FIG. 8 is an operation explanatory diagram of an effective displacement determination unit and a smoothing processing unit.

FIG. 9 is an operation explanatory diagram of an associating unit.

FIG. 10 is an operation explanatory diagram of a problem in the conventional image target detecting apparatus.

FIG. 11 is a configuration diagram of an image target detecting apparatus according to a conventional technique.

[Explanation of symbols]

1 Imager, 2 Point Image Enhancement Filter, 3 Position Measuring Unit, 4 Position Storage Unit, 5 Matching Unit, 14 Point Image Removal Filter, 15 First Image Storage Unit, 16 Second Image Storage Unit, 17 First Read control unit, 18 second read control unit, 19 image correlation calculation unit, 20 effective displacement determination unit, 21 smoothing processing unit, 22 moving target detection unit, 2
6 Subtraction unit.

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01B 11/00 G06T 1/00 G06T 7/20

Claims (6)

(57) [Claims] 1. An image pickup device for outputting an image signal, a point image enhancement filter for inputting an image signal output by the image pickup device for enhancing a point image included in the image signal, and an image signal output by the image pickup device. A point image removal filter that removes a point image included in an image signal as an input, a position measurement unit that obtains the position of the point image on the screen based on the output of the point image enhancement filter, and a position storage that stores the position of the point image From the point image removal filter , and a correspondence section that associates the point image detected on one screen with the point image detected on the next screen.
A first image storage unit for storing an image obtained by removing the image points, and a second image storage unit for storing an image picture obtained by removing the point from the point image removing filter, a rectangular area of the screen is divided into rectangular A rectangular area whose position on the screen is the same as that of another rectangular area including the rectangular area read by the first read control section. A place where a rectangular area output from the second read control unit and a rectangular area output from the first read control unit match in the rectangular area output from the second read control unit is determined. An image correlation calculation unit, an effective displacement determination unit that determines effective data from the displacements of the two screens output by the image correlation calculation unit, a smoothing processing unit that smoothes the effective displacement, a position of the point image and the smoothness. Output of processing unit And a moving target detection unit that detects a moving target by receiving the displacement of the image, the first image storage section, after temporarily storing the image, before
Until the moving target detection unit determines the moving target
The second image storage unit continues to hold the stored image,
Every time the image of the next screen is input, it will be memorized as the image of the new screen.
An image target detection device characterized by updating the contents . 2. An image pickup device that outputs an image signal, a point image removal filter that receives an image signal output from the image pickup device and removes a point image included in the image signal, and an image signal output from the image pickup device. A subtraction unit that subtracts the point image removed image, a position measurement unit that obtains the position of the point image on the screen based on the output of the subtraction unit, a position storage unit that stores the position of the point image, and a position detection unit that is detected on a certain screen. An associating unit for associating the point image with the point image detected on the next screen, and removing the point image
A first image storage unit that stores the image from which the point image from the filter is removed, a second image storage unit that stores the image from which the point image from the point image removal filter is stored, and a screen divided into rectangles. A rectangle whose position on the screen is the same as that of a first read control unit that reads a rectangular region from the first image storage unit and another rectangular region that includes the rectangular region read by the first read control unit. A second read control unit that reads an area from the second image storage unit and a rectangular area output by the first read control unit are coincident with each other in a rectangular area output by the second read control unit. An image correlation calculating unit for determining, an effective displacement determining unit for determining effective data from displacements of two screens output from the image correlation calculating unit, a smoothing processing unit for smoothing the effective displacement, and a position of a point image The smoothing process And a moving target detection unit outputting undergoing displacement of images for detecting a moving target, said first image storage section, after temporarily storing the image, before
Until the moving target detection unit determines the moving target
The second image storage unit continues to hold the stored image,
Every time the image of the next screen is input, it will be memorized as the image of the new screen.
An image target detection device characterized by updating the contents . 3. The effective displacement determination unit obtains a median value of displacements of a plurality of rectangular regions, and determines that a difference from the median value is within a predetermined range as a determination criterion. 2. The image target detecting device according to 2. 4. The smoothing processing unit obtains an average value of effective displacements of a plurality of rectangular areas, and when the number of effective displacements of the plurality of rectangular areas is small, the average of displacements around the rectangular area is used as the rectangular area. The displacement is defined as
Alternatively, the image target detecting device described in 2. 5. The position measuring unit binarizes the point image-enhanced image, and outputs a predetermined number of positions of the center of gravity of the regions in the connected region of the binary image in descending order of the maximum value of the luminance in the region. The image target detecting device according to claim 1, wherein the image target detecting device is provided. 6. The position measuring unit is configured to output the positions of a predetermined number of local maximum points in the descending order of the brightness of the local maximum points among the local maximum points of the point image enhanced image. 2. The image target detecting device according to 2.