JP2010014482A - Image target detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an image target detection device capable of stably performing target detection even when a clear sky background space has gradation.
A window circuit that cuts out luminance data from two regions sandwiching a target pixel in the raster scanning direction, an average luminance calculation circuit that calculates an average luminance of luminance data of the two regions cut out by the window circuit, and a captured image From the subtraction circuit for subtracting the average luminance calculated by the average luminance calculation circuit from the luminance data subtracted by the subtraction circuit, the average luminance and the standard deviation around the target pixel are calculated, and 2 is calculated from the calculated average luminance and standard deviation. An image target detection apparatus comprising: a threshold calculation circuit for calculating a binarization threshold; and a binarization circuit for binarizing luminance data subtracted by a subtraction circuit based on the binarization threshold calculated by the threshold calculation circuit Configure.
[Selection] Figure 1

Description

  The present invention relates to an image target detection apparatus that detects a target based on a photographed image.

  Conventionally, the luminance value of the target pixel and the noise luminance distribution in the vicinity of the target pixel are measured for the frame image obtained by capturing the target with the infrared imaging device, and binary using the measured value of the noise luminance distribution for each target pixel. By calculating the threshold value and binarizing the image captured using the calculated threshold value, it suppresses clutter such as clouds in the complex luminance distribution area and stabilizes the target in the clear sky area. An image target detection device that detects the image is known (for example, Patent Document 1). In this image target detection apparatus, the target labeling process is performed using the binarized image, and the center of gravity, area, and area are calculated as the target feature amount, so that a minute target such as an aircraft or a flying object flying over the sky. The minute target can be detected by using the image captured from the image.

Japanese Patent Laid-Open No. 6-121317

FIG. 8 is a diagram illustrating the relationship between the altitude captured by the infrared imaging device and the brightness of the clear sky background space.
The conventional image target detection device is based on the premise that the luminance distribution of noise in the clear sky background space of the minute target imaged by the infrared imaging device is a uniform distribution mainly due to the shot noise output by the infrared imaging device. , Stable target detection can be performed. However, in reality, as shown in FIG. 8, due to the influence of the atmospheric transmittance on the altitude, in the clear sky background space, the higher the altitude, the larger the luminance value, and the lower the altitude, the smaller the luminance value. It is known that brightness has gradation with respect to altitude.

  For this reason, when measuring the noise luminance distribution using the average luminance calculation circuit or the standard deviation calculation circuit, the measurement value of the average luminance calculation circuit varies depending on the gradation of the clear sky background, and the measurement value of the standard deviation calculation circuit However, the threshold value measured by the threshold value calculation circuit is extremely larger and fluctuates than the shot noise distribution output from the infrared imaging device. As a result, even if the luminance level corresponding to the target is higher than the shot noise output from the infrared imaging device, there is a problem that target detection cannot be performed sufficiently and stably.

  The present invention has been made in order to solve such problems, and an object thereof is to obtain an image target detection apparatus capable of stably performing target detection even when the clear sky background space has gradation. .

  The image target detection device according to the present invention is rotatably supported by a gimbal mechanism, and sandwiches a pixel of interest in the raster scanning direction based on a captured image of an infrared imaging device that is scanned while keeping the line-of-sight direction horizontal. A window circuit that cuts out luminance data from two areas, an average luminance calculation circuit that calculates the average luminance of the luminance data of the two areas cut out by the window circuit, and an average luminance calculation circuit that calculates from the captured image of the infrared imaging device The average luminance and standard deviation around the target pixel are calculated from the subtraction circuit that subtracts the average luminance and the luminance data subtracted by the subtraction circuit, and the binarization threshold is calculated from the calculated average luminance and standard deviation. 2 that binarizes the luminance data subtracted by the subtraction circuit based on the threshold calculation circuit and the binarization threshold calculated by the threshold calculation circuit And circuit, in which with a.

  A database storing rotational angle information of a background image in a frame image of the infrared imaging device measured in advance corresponding to a directivity direction of the infrared imaging device rotatably supported by the gimbal mechanism; Based on the orientation direction of the infrared imaging device obtained by the drive control unit, rotation angle information is obtained from the database, and based on the obtained rotation angle information, the direction of the background gradation matches the direction perpendicular to the raster scan. An image correction circuit for performing image correction by reversely rotating the background image, and a window circuit for cutting out luminance data from two regions sandwiching the target pixel in the raster scanning direction based on the image corrected by the image correction circuit; An average luminance calculation circuit for calculating the average luminance of the luminance data of the two regions cut out by the window circuit; A subtraction circuit that subtracts the average luminance calculated by the average luminance calculation circuit from the image captured by the infrared imaging device, and the average luminance and standard deviation around the target pixel are calculated from the luminance data subtracted by the subtraction circuit. A threshold value calculation circuit for calculating a binarization threshold value from the averaged luminance and standard deviation, and binarization for binarizing the luminance data subtracted by the subtraction circuit based on the binarization threshold value calculated by the threshold value calculation circuit. And a circuit.

  According to the present invention, it is possible to detect a target sufficiently stably by detecting a difference in signal level between the target and a uniform peripheral region without being affected by the background gradation.

Embodiments of the present invention will be described below with reference to the drawings. In addition, in each embodiment, the same code | symbol is attached | subjected to the same structure, and duplication description is abbreviate | omitted.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing the configuration of an image target detection apparatus according to Embodiment 1 of the present invention.
In the figure, an infrared imaging device 1 is stored in a gimbal mechanism unit 10 and mounted on a body or the like flying over the altitude, and an infrared detector is used to capture an infrared image. The gimbal mechanism 10 pivotally supports the infrared imaging device 1. The gimbal drive control unit 11 rotationally drives the gimbal mechanism unit 10 around two rotational axes in the vertical direction and the azimuth direction, and directs the line-of-sight direction of the infrared imaging device 1 to an arbitrary altitude and azimuth, whereby the infrared imaging device 1 Images a target such as a flying object or aircraft flying over the sky. The gimbal drive control unit 11 rotates the gimbal mechanism unit 10 so that the line-of-sight direction of the infrared imaging device 1 is kept horizontal, and scans an image captured by the infrared imaging device 1. The image target detection device 9a receives an image captured by the infrared imaging device 1, and automatically detects a target for the input image.

  Note that the gimbal drive control unit 11 rotates the gimbal mechanism unit 10 about the visual axis of the infrared imaging device 1 so that the image captured by the infrared imaging device 1 does not rotate about the visual axis of the infrared imaging device 1. In addition, the gimbal mechanism 10 may be configured.

The image target detection device 9 a includes a preprocessing unit 15 that uniformly suppresses background gradation, a target detection unit 8, and a feature amount calculation circuit 7.
The preprocessing unit 15 includes a window circuit 12, an average luminance calculation circuit 14, and a subtraction circuit 13. The window circuit 12 is a pixel in two fixed areas in the vicinity of each pixel (hereinafter referred to as a target pixel) constituting the frame image in the raster scanning direction with respect to the frame image output from the infrared imaging device 1. Cut out the brightness data. The average luminance calculation circuit 14 calculates an average luminance value in the luminance data of the pixels cut out by the window circuit 12. The subtraction circuit 13 subtracts the average luminance value measured by the average luminance calculation circuit 14 from the luminance data of the target pixel.

  The target detection unit 8 includes a window circuit 2, an average luminance calculation circuit 3, a standard deviation calculation circuit 4, a threshold value calculation circuit 5, and a spatial binarization circuit 6. The window circuit 2 cuts out pixel data in a window around the pixel for each pixel of interest with respect to the luminance data of the frame image subtracted by the subtraction circuit 13 of the preprocessing unit 15. The average luminance calculation circuit 3 measures the average luminance value in the window around the pixel of interest for the pixel data cut out by the window circuit 2. The standard deviation calculation circuit 4 measures the standard deviation value in the window around the target pixel for the pixel data cut out by the window circuit 2 and the luminance average value measured by the average luminance calculation circuit 3.

  The threshold calculation circuit 5 binarizes the pixel of interest using the luminance average value, standard deviation value, and Equation 1 in the window near the pixel of interest measured by the average luminance calculation circuit 3 and the standard deviation calculation circuit 4. Perform threshold calculation.

  The spatial binarization circuit 6 performs binarization processing on the luminance data of the frame image subtracted by the subtraction circuit 13 of the preprocessing unit 15 using the binarization threshold obtained by the threshold calculation circuit 5. . The spatial binarization circuit 6 calculates a binarized image using the binarization threshold value for each pixel of interest, the luminance value of the pixel of interest, and Equation 2, and the binarization threshold as shown in Equation 2 above. Target detection is performed by regarding a pixel obtained by binarizing a pixel of interest with higher luminance into “1” as a target.

FIG. 2 is a diagram illustrating a calculation principle of binarization processing in the spatial binarization circuit 6 of the target detection unit 8.
For the frame image input in real time from the pre-processing unit 15, the average luminance calculation circuit 3 and the standard deviation calculation circuit 4 perform the luminance average of noise in the vicinity of the target pixel in real time while performing raster scanning in the target detection unit 8. By measuring the value and the luminance standard deviation value, the threshold value calculation circuit 5 can set a threshold value for each target pixel according to the luminance distribution of noise near the target pixel. For this reason, the binarization threshold value which makes the target false alarm rate always constant can be set for each target pixel.

  The feature amount calculation circuit 7 performs a target labeling process using the binarized image processed by the spatial binarization circuit 6 of the target detection unit 8, calculates the center of gravity, area, and region as the target feature amount, The result is output. As described above, the image target detection device 9 suppresses clutter such as clouds in a complex luminance distribution region from an image obtained by imaging a minute target such as an aircraft or a flying object from above, and stabilizes the target. Can be detected.

Next, the operation of the image target detection device 9a will be described.
The gimbal drive control unit 11 operates so as to keep the azimuth angle of the gimbal mechanism unit 10 horizontal. As a result, the gimbal drive control unit 11 physically matches the direction of the background gradation in the image captured by the infrared imaging device 1 with the direction perpendicular to the raster scan, as shown in FIG. The window circuit 12 cuts out image data in two regions 101 that sandwich the target pixel 100 in the vicinity of the vicinity of the target pixel 100.

  Here, since the image data cut out by the window circuit 12 is composed of luminance data in the line direction perpendicular to the gradation of the background luminance 102, uniform luminance data independent of the gradation can be obtained.

  Further, the average luminance calculation circuit 14 calculates the average value of the image data cut out by the window circuit 12, and the subtraction circuit 13 subtracts the average value of the image data calculated by the average luminance calculation circuit 14 from the luminance value of the target pixel. To do.

  As a result, the luminance data output from the preprocessing unit 15 can detect a difference in signal level between uniform peripheral areas that do not depend on the target pixel and gradation, and without deteriorating the luminance level corresponding to the target. Background gradation can be suppressed uniformly. Thus, since the target detection unit 8 can use a uniform image in which the background gradation is suppressed, the target can be detected stably without being affected by the background gradation.

  As described above, in the image target detection device according to the first embodiment, the gimbal mechanism unit incorporating the infrared imaging device mounted on the aircraft flying above the altitude is gimbaled while keeping the line of sight of the infrared imaging device horizontal. By performing scanning imaging to control the vertical angle and azimuth angle of the mechanism unit, the direction of the gradation in the background generated by the influence of atmospheric transmittance, etc. on the altitude in the captured image of the infrared imaging device is rasterized in the image target detection device. Match in the direction perpendicular to the scan. In this way, uniform luminance data independent of gradation is obtained by the window circuit that cuts out the two fixed regions in the vicinity of the target pixel in the raster scanning direction, the average value of the luminance data is calculated, and the above-mentioned luminance value of the target pixel is calculated. By including a pre-processing unit that subtracts the average value of the target, the target detection is sufficiently stable by detecting the difference between the signal level of the target and the uniform peripheral region without being affected by the background gradation. be able to.

Embodiment 2. FIG.
FIG. 4 is a block diagram showing a configuration of an image target detection device 9b according to Embodiment 2 of the present invention. The preprocessing unit 15a of the image target detection device 9b includes a frame memory 16, a subtraction circuit 13, a window circuit 12a, and an average luminance calculation circuit 14. The target detection unit 8 of the image target detection device 9b is the same as that in the first embodiment. Here, the pre-processing unit 15a of the image target detection device 9b replaces the window circuit 12 that cuts out the two fixed areas in the pre-processing unit 15 of the first embodiment with the window circuit 12b that cuts out the two variable areas, and performs infrared rays. A frame memory 16 for acquiring a frame image output from the imaging unit 1 is provided.

  The frame memory 16 stores a frame image obtained by the infrared imaging device 1. Based on the captured image stored in the frame memory 16, the window circuit 12a varies the size of two variable areas sandwiching the target pixel in the raster scanning direction, and the pixel brightness data from the two variable areas is variable. Cut out. The subtracting circuit 13 subtracts the average luminance calculated by the average luminance calculating circuit 14 from the frame image stored in the frame memory 16.

Next, the operation of the image target detection device 9b will be described.
Depending on the operation of the apparatus, the target size in the frame image captured by the infrared imaging apparatus 1 is not necessarily constant. For this reason, in the pre-processing unit 15a, the gap between the two windows arranged in the raster scanning direction constituting the variable window circuit 12a is matched with the target size assumed on the frame screen output from the infrared imaging device 1, Allow operators to make variable settings. As a result, a target having a size that needs to be detected by operation can be stably detected without being affected by the background gradation.

  After a part of the image area is cut out from the frame image output from the infrared imaging device 1 by the window circuit 12a, the average luminance calculation circuit 14 performs calculation processing. For this reason, the noise variation (standard deviation) superimposed on the image data after preprocessing by the preprocessing unit 15a is statistically different from the noise variation (standard deviation) superimposed on the frame image captured by the infrared imaging device 1. It is generally known that the noise variation (standard deviation) after pre-processing by the pre-processing unit 15a increases as the upper (1 + 1 / m) times increase and m decreases.

  Here, m is the number of pixels cut out by the window circuit 12a. Conversely, as m is increased, there is a concern that in the frame image output from the infrared imaging device 1, low-frequency noise superimposed on the line direction perpendicular to the gradation is amplified. Therefore, the frame image output from the infrared imaging device 1 is acquired in the frame memory 16, the noise distribution superimposed on the acquired frame image is analyzed by the operator, and the window cut out by the window circuit 12a according to the analyzed noise distribution. By enabling the number of pixels in the region to be adjusted, noise variation (standard deviation) after processing by the preprocessing unit 15a is suppressed. Thereby, an increase in noise variation (standard deviation) by the preprocessing unit 15a can be prevented, and the target detection unit 8 in the subsequent stage can stably detect the target.

  As described above, the image target detection apparatus according to the second embodiment includes the frame memory 16 that acquires the frame image output from the infrared imaging apparatus. Further, in the window circuit 12a, the variable two regions near both the target pixels in the raster scanning direction are matched with the target size assumed in operation on the frame screen output from the infrared imaging device 1 and stored in the frame memory 16. The operator cuts out the pixel data by changing the size of the variable 2 area. As a result, a target having a size that needs to be detected in operation can be detected stably without being affected by the background gradation.

  In addition, since the operator adjusts the number of pixels in the image area cut out in the window circuit 12a according to the noise distribution superimposed on the acquired frame image, uniform luminance data independent of gradation can be obtained. Then, the average value of the luminance data obtained by the average luminance calculation circuit 14 is calculated, and the average value calculated from the luminance value of the target pixel is subtracted by the subtraction circuit 15a. Thus, an increase in noise variation (standard deviation) caused by the action of the preprocessing unit 15a can be prevented, and the target detection can be stably performed by the target detection unit 8.

Embodiment 3 FIG.
FIG. 5 is a block diagram showing a configuration of an image target detection apparatus 9c according to Embodiment 3 of the present invention. In the figure, a gimbal mechanism unit 10 rotatably supports an infrared imaging device 1 that captures an infrared image by an infrared detector. The gimbal drive control unit 11 rotationally drives the gimbal mechanism unit 10 to control the direction of the line of sight of the infrared imaging device 1 in a predetermined direction, and the altitude information (vertical angle, elevation angle, or depression angle) of the line of sight of the infrared imaging device 1. ) And azimuth information (azimuth angle) are output to the image target detection device 9c.

  The image target detection device 9c according to the third embodiment is configured by adding an image correction circuit 17 to the image target detection device 9b according to the second embodiment shown in FIG. The frame memory 16 stores a frame image obtained by the infrared imaging device 1. The image correction circuit 17 performs geometric image rotation processing based on altitude information and azimuth angle information from the gimbal drive control unit 11 from the image data read from the frame memory 16.

Next, the operation of the image correction circuit 17 will be described.
The gimbal mechanism unit 10 supports the infrared imaging device 1 around two rotation axes. The gimbal mechanism unit 10 directs the line of sight of the infrared imaging device 1 in a desired direction around two rotation axes based on arbitrary altitude information and azimuth information by the gimbal drive control unit 11, and a flying object, an aircraft, etc. flying over the sky The target is imaged and scanned. FIG. 6 is a view showing the principle of background image rotation by altitude and direction orientation from the gimbal drive control unit 11. As shown in FIG. 6, an image captured by the infrared imaging device 1 supported by the gimbal mechanism unit 10 is generally rotated by a captured background image due to, for example, the action of a coupe optical system constituting the infrared imaging device 1. Known. As a result of this rotating action, the background gradation direction and the raster scanning direction are not necessarily perpendicular to the image captured by the infrared imaging device 1. For this reason, the image data cut out by the window circuit 12a cannot always be composed of luminance data in the line direction perpendicular to the gradation, and the luminance data output from the preprocessing unit 15 is influenced by the background gradation. The received nonuniform luminance data.

Therefore, the image target detection apparatus 9c according to the third embodiment performs the following processing so that the image data cut out by the window circuit 12a is composed of luminance data in the line direction perpendicular to the gradation.
First, the frame image captured by the infrared imaging device 1 is temporarily stored in the frame memory 16. In the image correction circuit 17, the altitude and direction directed by the gimbal drive control unit 11 and the rotation angle of the background image in the frame image captured by the infrared imaging device 1 are measured in advance and stored as a database. To do. This database may be provided inside the image correction circuit 17 or may be provided in an external device of the image correction circuit 17.

  Next, in synchronization with the frame image captured by the infrared imaging device 1, the gimbal drive control unit 11 transmits the altitude and azimuth directed by the gimbal drive control unit 11 to the image correction circuit 16. In the image correction circuit 17, the rotation angle of the background image obtained by referring to the database from the altitude and azimuth directed by the gimbal drive control unit 11 with respect to the frame image captured by the infrared imaging device 1 is used. As shown in Equation 3, the background image is reversely rotated.

  Note that the position of the image data after the reverse image rotation does not necessarily geometrically coincide with the position of the original pixel that constitutes the detector of the infrared imaging device 1. For this reason, in the image data before the reverse image rotation shown in FIG. 9, the image data at the four pixel positions that exist in the infrared detector adjacent to the periphery of the target pixel after the reverse image rotation is used, Perform interpolation processing.

  Thus, the image correction circuit 17 can electrically match the background gradation direction in the image captured by the infrared imaging device 1 with the direction perpendicular to the raster scan, and the image data cut out by the window circuit 12a is Consists of luminance data in the line direction perpendicular to the gradation. For this reason, uniform luminance data independent of gradation can be obtained.

  The average luminance calculation circuit 14 calculates the average value of the image data cut out by the window circuit 12a, and the subtraction circuit 13 subtracts the average value from the luminance value of the target pixel. As a result, the luminance data output from the preprocessing unit 15 can detect a difference in signal level between the target pixel and a uniform peripheral region that does not depend on the gradation, and the background can be detected without degrading the luminance level corresponding to the target. Can be suppressed uniformly. At this time, since the target detection unit 8 can use an image in which the background gradation is suppressed, even if the gimbal mechanism unit 10 is directed to any altitude and direction, the target detection unit 8 is stable without being affected by the background gradation. Target detection.

  As described above, in the image target detection device 9c of the third embodiment, the captured image of the infrared imaging device 1 that is scanned and imaged with the gimbal mechanism unit 10 incorporating the infrared imaging device 1 oriented at an arbitrary altitude and direction. On the other hand, a database in which the altitude and azimuth directed by the gimbal drive control unit 11 and the rotation angle of the background image in the frame image captured by the infrared imaging device 1 with respect to the altitude and orientation is previously measured and stored is provided. Further, using the altitude and azimuth directed by the gimbal drive control unit 11, the background image is reversely rotated so that the direction of the background gradation is electrically coincident with the direction perpendicular to the raster scan in the image target detection device 1, In addition, an image correction circuit 17 that interpolates the image data after the image rotation using the data of the adjacent four pixels after the actual image reverse rotation is provided. As a result, even if the gimbal mechanism unit 10 is directed to any altitude / azimuth, the target can be detected stably without being affected by the background gradation.

  Thus, in the image target detection apparatus according to the third embodiment, the gimbal drive control section 11 keeps the azimuth of the gimbal mechanism section 10 required in the image target detection apparatuses 9a and 9b according to the first and second embodiments. Thus, there is no need for operational restrictions to physically match the background gradation direction in the image captured by the infrared imaging device 1 with the direction perpendicular to the raster scan. For this reason, the target detection unit 8 can stably detect the target in correspondence with any altitude / azimuth orientation of the gimbal mechanism unit 10.

It is a block diagram which shows the structure of the image target detection apparatus by Embodiment 1 of this invention. It is a figure which shows the calculation principle of the binarization threshold value of an image target detection apparatus. It is a figure which shows the relationship between the direction of the background gradation which the infrared imaging device by Embodiment 1 of this invention images, and a raster scanning direction. It is a block diagram which shows the structure of the image target detection apparatus by Embodiment 2 of this invention. It is a block diagram which shows the structure of the image target detection apparatus by Embodiment 3 of this invention. It is a figure which shows the principle of the background image rotation by an altitude and direction orientation from a gimbal drive control part. It is a figure which shows the interpolation process in the image correction circuit of the image target detection apparatus by Embodiment 3 of this invention. It is a figure explaining the relationship between the altitude which an infrared imaging device images, and the brightness | luminance of clear sky background space.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Infrared imaging device, 2 window circuit, 3 average brightness calculation circuit, 4 standard deviation calculation circuit, 5 threshold value calculation circuit, 6 space binarization circuit, 7 feature-value calculation circuit, 8 target detection part, 9a, 9b, 9c Image target detection device, 10 Gimbal mechanism unit, 11 Gimbal drive control unit, 12, 12a Window circuit, 13 Subtraction circuit, 14 Average luminance calculation circuit, 15 Preprocessing unit, 16 Frame memory, 17 Image correction circuit

Claims (3)

Based on a captured image obtained by an infrared imaging device that is pivotally supported by a gimbal mechanism and scanned while keeping the line-of-sight direction horizontal, luminance data is obtained from two regions sandwiching the target pixel in the raster scanning direction. A window circuit to cut out,
An average luminance calculation circuit for calculating the average luminance of the luminance data of the two regions cut out by the window circuit;
A subtracting circuit for subtracting the average luminance calculated by the average luminance calculating circuit from the captured image of the infrared imaging device;
A threshold value calculation circuit for calculating an average luminance and a standard deviation around the target pixel from the luminance data subtracted by the subtraction circuit, and calculating a binarization threshold value from the calculated average luminance and the standard deviation;
A binarization circuit for binarizing the luminance data subtracted by the subtraction circuit based on the binarization threshold calculated by the threshold calculation circuit;
An image target detection apparatus. A frame memory for storing a frame image taken by the infrared imaging device;
The window circuit varies the size of the two regions sandwiching the pixel of interest in the raster scanning direction based on the frame image stored in the frame memory, and extracts luminance data from the two variable regions,
2. The image target detecting apparatus according to claim 1, wherein the subtracting circuit subtracts the average luminance calculated by the average luminance calculating circuit from a photographed image stored in the frame memory. A database storing rotation angle information of a background image in a frame image of the infrared imaging device measured in advance, corresponding to the directivity direction of the infrared imaging device rotatably supported by the gimbal mechanism,
Rotation angle information is obtained from the database based on the orientation direction of the infrared imaging device, and the background image is reversely rotated based on the obtained rotation angle information so that the direction of the background gradation coincides with the direction perpendicular to the raster scan. An image correction circuit for correcting the image,
A window circuit that cuts out luminance data from two regions sandwiching the pixel of interest in the raster scanning direction based on the image corrected by the image correction circuit;
An average luminance calculation circuit for calculating the average luminance of the luminance data of the two regions cut out by the window circuit;
A subtracting circuit for subtracting the average luminance calculated by the average luminance calculating circuit from the captured image of the infrared imaging device;
A threshold calculation circuit for calculating an average luminance and a standard deviation around the target pixel from the luminance data subtracted by the subtraction circuit, and calculating a binarization threshold from the calculated average luminance and the standard deviation;
A binarization circuit for binarizing the luminance data subtracted by the subtraction circuit based on the binarization threshold calculated by the threshold calculation circuit;
An image target detection apparatus.

Images