JP2001056860A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the erroneous detection of a background area by calculating a luminance value to become an extremal value by calculating the ratio of numbers of significant pixels for each luminance value from the number of significant pixels when a measured image is binarized for each luminance value, and defining the luminance value of the extreme value close to '1' as a threshold value. SOLUTION: A first histogram measuring circuit 8a measures the frequency of generation for each luminance value of the measured image, a low-pass filter circuit 20 extracts the low frequency component of the measured value and outputs it as a low frequency component image and a second histogram measuring circuit 8b measures the frequency of generation for each luminance value of a low frequency component image 21. A significant pixel number ratio calculating circuit 22 calculates the ratio of numbers of significant pixels for each luminance value from the number of significant pixels when the measured value of the first histogram measuring circuit 8a is inputted and a measured image 7 is binarized for each luminance value and the number of significant pixels when the measured value of the second histogram measuring circuit 8b is inputted and the measured image 7 is binarized for each luminance value. A first threshold value selecting circuit 23 calculates the luminance value having the extreme value of the significant pixel number ratio and selects the luminance value of the extreme value close to '1' as a threshold value 10.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus mounted on an aircraft or a ship to detect a target of the ship or the aircraft using an infrared image.

[0002]

2. Description of the Related Art First, an image processing apparatus of this type will be described with reference to FIG. In FIG. 9, reference numeral 1 denotes an aircraft equipped with this type of image processing device, reference numeral 2 denotes an image processing device, and reference numeral 3 denotes a target, for example, a ship running on the sea. FIG. 10 is a block diagram showing the configuration of the image processing device 2 in FIG. 9, and in FIG. 10, reference numeral 4 denotes an infrared imaging circuit that outputs a voltage value that changes according to the intensity of infrared radiation as a grayscale image for each pixel; Reference numeral 5 denotes a gray-scale image, 6 denotes a gate setting circuit that sets a measurement area in the gray-scale image 5 and outputs the measurement image, 7 denotes a measurement image, and 8a denotes a first histogram measurement circuit that measures an occurrence frequency for each luminance value of the measurement image. , 9 is a threshold value calculation circuit for calculating a threshold value from the measurement result of the first histogram measurement circuit 8a, 10 is a threshold value, 11a is a first value for comparing the luminance value of the measurement image 7 with the threshold value 10 and outputting a binary image. 12 is a binary image, 13 is a barycentric value calculating circuit that determines the connection state of significant pixels of the binary image 12 and calculates the barycentric position,
Reference numeral 14 denotes a center of gravity value calculated by the center of gravity value calculation circuit 13.

The conventional image processing apparatus is configured as described above, and the infrared imaging circuit 4 outputs a voltage value that changes according to the radiation intensity of the infrared light as a grayscale image 5 for each pixel. The gate setting circuit 6 sets a measurement area in the grayscale image 5 and outputs a measurement image 7 in which the processing area is limited. The first histogram measurement circuit 8 measures the histogram of the measurement image 7.

When the threshold value calculating circuit 9 divides the histogram measurement value into two classes of a significant region and a background region, it sets a threshold value 10 so that the average value between the classes is as large as possible and the variance within the class is as small as possible. calculate. (1
IEICE published in 980 by IEICE
(D), J63-D, pp. 346-356, "Automatic threshold decision method based on discrimination and least square criterion". )

The first binarizing circuit 11a binarizes the measured image 7 with a threshold value 10 and outputs a binary image 12 by an operation shown in, for example, "Equation 1".

[0006]

(Equation 1)

Here, B (I, J) is the luminance value of the measurement image 7 at the position (I, J), Thresh is the threshold value 10, and Bin1 (I, J) is the binary image 12.

The first binarization circuit 11a outputs, for example, "1" as a significant pixel if the luminance value of the measured image 7 is larger than the threshold value 12, and outputs "0" as a small pixel is not significant. . The center-of-gravity value calculation circuit 13 calculates the binary image 1
2, the connected state of each significant pixel is determined, and the center of gravity value 14 for each connected area is output as a target position.

[0009]

Since the conventional apparatus is configured as described above, as shown in FIG. 9, when the target 3 is mounted on the aircraft 1 and imaged by the infrared imaging circuit 5, the threshold value is calculated. When the circuit 9 divides the histogram measured by the first histogram calculation circuit 8a into two classes, an object and a background, it calculates a threshold value so that the average value between the classes is as large as possible and the variance within the class is as small as possible. Therefore, if there is a difference between the number of pixels in the background area and the number of pixels in the target area, the threshold value 10 is set to a luminance value forming the background area.

FIG. 11 shows an example of the gray image 5 obtained in FIG. Reference numeral 15 denotes a measurement area set by the gate setting circuit 6. 16 is a target area, and 17 is a background area.

FIG. 12 shows a first histogram measuring circuit 8a.
2 shows a measurement example. Reference numeral 18 denotes an occurrence frequency distribution of the background area 17, and 19 denotes an example of the occurrence frequency distribution of the target area 16.

When there is a difference between the number of pixels in the target area and the number of pixels in the background area, the threshold value calculating circuit 9 calculates the threshold value 10 so as to reduce the luminance variance of the background pixels by the action of reducing the variance in the class. . For example, a point A as shown in FIG.

Therefore, there is a problem that a significant part of the binary image 12 includes a part of the background area 17 and a part of the background area 17 is erroneously detected as a target.

The present invention has been made to solve such a problem, and an object of the present invention is to provide an image processing apparatus capable of preventing erroneous detection of a background area.

[0015]

According to a first aspect of the present invention, there is provided an image processing apparatus comprising: first histogram measuring means for measuring an occurrence frequency for each luminance value of a measured image; A low-frequency component image extracting unit that outputs a low-frequency component image, a second histogram measuring unit that measures the frequency of occurrence for each luminance value of the low-frequency component image, and a measurement value of the first histogram measurement unit. The number of significant pixels when the measurement image is binarized for each luminance value and the measurement value of the second histogram measurement unit are input, and the number of significant pixels when the measurement image is binarized for each luminance value And a significant pixel number ratio calculating means for calculating a significant pixel number ratio for each brightness value, calculating a brightness value at which the significant pixel number ratio is an extreme value, and selecting a brightness value at which the extreme value is close to 1 as a threshold value And a threshold selecting means.

An image processing apparatus according to a second aspect of the present invention includes a first histogram measuring means for measuring an occurrence frequency for each luminance value of a measurement image, and a low-frequency component of the measurement image, which is output as a low-frequency component image. A low-frequency component image extracting means, a second histogram measuring means for measuring an occurrence frequency for each luminance value of the low-frequency component image, and a measurement value of the first histogram measuring means, and The number of significant pixels when binarized for each value and the measurement value of the second histogram measuring unit are input, and the number of significant pixels when the measured image is binarized for each luminance value is calculated for each luminance value. A significant pixel number ratio calculating means for calculating a significant pixel number ratio, comparing the parameter value with the extreme value of the significant pixel number ratio, and not exceeding the parameter value;
And a threshold value selecting means for selecting an extreme luminance value close to 1 as a threshold value.

An image processing apparatus according to a third aspect of the present invention has a first histogram measuring means for measuring the frequency of occurrence for each luminance value of a measured image, and extracts a low-frequency component of the measured image and outputs it as a low-frequency component image. A low-frequency component image extracting means, a second histogram measuring means for measuring an occurrence frequency for each luminance value of the low-frequency component image, and a measurement value of the first histogram measuring means, and The number of background pixels when binarized for each value and the measurement value of the second histogram measuring means are input, and the number of background pixels when the measurement image is binarized for each luminance value is calculated for each luminance value. A background pixel number ratio calculating unit that calculates a background pixel number ratio; and a threshold value selecting unit that calculates a luminance value at which the background pixel number ratio is an extreme value and selects a luminance value whose extreme value is close to 1 as a threshold value. It is a thing.

An image processing apparatus according to a fourth aspect of the present invention includes a first histogram measuring means for measuring an occurrence frequency for each luminance value of a measured image, and a low frequency component of the measured image is extracted and output as a low-frequency component image. A low-frequency component image extracting means, a second histogram measuring means for measuring an occurrence frequency for each luminance value of the low-frequency component image, and a measurement value of the first histogram measuring means, and The number of background pixels when binarized for each value and the measurement value of the second histogram measuring means are input, and the number of background pixels when the measurement image is binarized for each luminance value is calculated for each luminance value. Background pixel number ratio calculating means for calculating the background pixel number ratio, comparing the parameter value with the extreme value of the background pixel number ratio, and does not exceed the parameter value;
And a threshold value selecting means for selecting an extreme luminance value close to 1 as a threshold value.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 FIG. 1 is a configuration diagram showing the first embodiment. In FIG. 1, reference numeral 20 denotes a low-pass filter circuit for extracting a low-frequency component of the measurement image and outputting the extracted low-frequency component as a low-pass component image; 21, a low-pass component image; The second histogram measurement circuit 8a, the first histogram measurement circuit 8a
When the measured image is input and the measured image 7 is binarized for each luminance value, and the number of significant pixels and the measured value of the second histogram measuring circuit 8b are input and the measured image 7 is binarized for each luminance value A significant pixel number ratio calculating circuit that calculates a significant pixel number ratio for each luminance value from the significant pixel number of the luminance value; and 23 calculates a luminance value at which the significant pixel number ratio has a maximum value, and a luminance value at which the maximum value is close to 1 To threshold 1
It is a first threshold value selection circuit that is selected as 0. 4 to 7, 8a, 10, 11a and 12 to 14 are equivalent to those in FIG.

The significant pixel number ratio calculating circuit 22 first converts the number of significant pixels and the low-frequency component image 21 when the measured image 7 is binarized for each luminance value by using “Expression 2” and “Expression 3”. And the number of significant pixels when binarized.

[0021]

(Equation 2)

[0022]

(Equation 3)

SUM_all_w (B) where “Expression 2” indicates that the measured image 7 is binarized by the luminance value B.

A significant pixel number ratio calculating circuit 22 calculates a significant pixel number ratio when the measured image 7 and the low-frequency component image 21 are respectively binarized at the same luminance value. “Equation 4” is an example of calculating the significant pixel number ratio.

[0025]

(Equation 4)

Est_w (B) where “Equation 4” represents the ratio of significant pixels.

FIG. 2 shows an example of a change in the ratio of the number of significant pixels in "Equation 4". 2A shows a luminance distribution state near the target area 16 of the measurement image 7 in FIG. 11, and FIG. 2B shows a luminance distribution state near the target area 16 of the low-frequency component image 21 in FIG. FIG. 2C shows a change in the significant pixel number ratio and the luminance value in “Equation 4”.

The measured image 7 with the threshold value within the range of FIG.
Is binarized, significant pixels are generated. On the other hand, in the low-frequency component image 21, high-frequency components due to noise have been removed, so that no effective pixels are generated even if binarization is performed using the same threshold value. Therefore, "Equation 3" becomes larger than "Equation 2", and the ratio of the number of significant pixels of "Equation 4" becomes larger than "1".

When the threshold value is within the range shown in FIG. 2A, the number of significant pixels when the low-frequency component image 21 is binarized increases, the ratio of the number of significant pixels according to "Equation 4" decreases, and the minimum point To form

When the threshold value is within the range of FIG. 2A, the low-frequency component image 21 is affected by the high-frequency component of the target edge portion.
Is smaller than the number of significant pixels of “Expression 2” in the measurement image, and the ratio of the number of significant pixels of “Expression 4” is increased.

When the threshold value is within the range shown in FIG. 2A, the number of significant pixels of the measured image and the low-frequency component image becomes equal, and the ratio of the number of significant pixels in "Equation 4" drops to around "1". .

When the threshold value is within the range shown in FIG. 2A, the noise component in the background region is a high-frequency component, so that the number of significant pixels of the measured image increases, and the ratio of the number of significant pixels in "Equation 4" increases. .

When the threshold value is within the range shown in FIG. 2A, the number of significant pixels is larger in the low-frequency component image, and the significant pixel number ratio of "Equation 4" is reduced to form a minimum point.

When the threshold value is within the range of FIG. 2A, the ratio of the number of significant pixels between the case where the low-frequency component image is binarized and the case where the measured component image is binarized becomes substantially the same value. The significant pixel number ratio of “4” is almost “1”.

The high-frequency component of the target edge of the measurement image 7 and the high-frequency component due to noise form the maximum point in the ratio of the number of significant pixels in “Equation 4”. In the luminance value B near the maximum point due to the high frequency component of the noise, many pixels that are not binarized in the low-frequency component image occur as significant pixels in the measurement image 7 and are larger than the significant pixel ratio of the maximum point due to the edge of the target area. Become.
Therefore, the first threshold value selection circuit 23 can calculate a threshold value at which the background region and the minute target region can be separated by selecting the local maximum point whose significant pixel ratio is close to “1”. For example, by selecting the threshold A shown in FIG. 2, the target position can be detected without erroneous detection.

Embodiment 2 FIG. 3 is a configuration diagram showing the second embodiment. In the figure, reference numeral 24 denotes a second threshold value selection circuit for selecting a luminance value close to "1" as a threshold value among the extreme values of the significant pixel number ratio that do not exceed the parameter value. 25 is a parameter value.

FIG. 4 shows an example of a change in the ratio of the number of significant pixels when the target area is not present in the measurement image 7. FIG.
4A shows the luminance distribution of the measurement image 7, FIG. 4B shows the luminance distribution of the low-frequency component image 21, and FIG. 4C shows the change in the ratio of the significant pixels according to "Equation 4".

Depending on the distribution of high frequency components due to noise in the background area, an extreme value may be formed at a significant pixel ratio as shown at point A in FIG. 4C.

In order to prevent erroneous detection in the measurement image 7 in which the target area does not exist, the parameter value 25 is compared with the extreme value of the ratio of the number of pixels, and the brightness of the extreme value which does not exceed the parameter value 25 and is close to "1" Choose the value as the threshold.

Embodiment 3 FIG. 5 is a configuration diagram showing the third embodiment. In FIG. 5, reference numeral 26 denotes the input of the measurement value of the first histogram measurement circuit 8a and the number of background pixels when the measurement image 7 is binarized for each luminance value and the measurement value of the second histogram measurement circuit 8b. A background pixel number ratio calculation circuit that calculates a background pixel number ratio for each luminance value from the number of background pixels when the measurement image is binarized for each luminance value, and 11b calculates the luminance value of the measurement image 7 and the threshold value 10 This is a second binarizing circuit that outputs a pixel having a luminance value lower than a threshold value as a binary image 12 in comparison.

The background pixel number ratio calculating circuit 26 calculates "Equation 5" from the measured histogram of the first histogram measuring circuit 8a.
And the number of background pixels when the measurement image is binarized for each luminance value B as shown in FIG. 8 and the low-frequency component image 2 as shown in “Expression 6” from the measurement histogram of the second histogram measurement circuit 8b.
The number of background pixels when 1 is binarized for each luminance value B is obtained.

[0042]

(Equation 5)

[0043]

(Equation 6)

SUM_all_b (B) where “Equation 5” indicates that the measured image 7 is binarized by the luminance value B

Next, the ratio of the number of background pixels when the measured image 7 and the low-frequency component image 21 are respectively binarized by the luminance value B is obtained. "Equation 7" is an example of calculating the background pixel number ratio.

[0046]

(Equation 7)

Est_b (B) Here, “Equation 7” represents the ratio of the number of background pixels.

"Equation 8" shows an operation example of the second binarization circuit 11b.

[0049]

(Equation 8)

However, Bin2 (I, J) represents the binary image 12 at the position (I, J).

FIG. 6 shows the change in the ratio of the number of background pixels in "Equation 7". FIG. 6A shows the measurement.

As shown in FIG. 6, the operation of the background pixel number ratio calculating circuit 26 is equivalent to that of the significant pixel number ratio calculating circuit 22, so that a local maximum occurs as in FIG. 2 (c). As in FIG. 2, FIG. 6 shows a local maximum point where the ratio of the number of background pixels is close to “1”.
By selecting the point A shown in (c), even when the luminance value of the target area is lower than the luminance value of the background area, a maximum point occurs due to the high frequency of the edge of the target area. It can be extracted stably.

Embodiment 4 FIG. 7 is a configuration diagram showing the fourth embodiment. 24 and 25 are the same as FIG. 3 and 26 are the same as FIG.

FIG. 8 shows the operation of the background pixel ratio calculating circuit 26 when the target area does not exist in the measurement image 7. Since the operation of the parameter value 25 is equivalent to that described in the second embodiment, the second threshold value selection circuit 24
Compares the parameter value 25 with the extreme value of the ratio of the number of background pixels, and selects a luminance value of the extreme value that does not exceed the parameter value 25 and is close to “1” as the threshold value.

Even when the background pixel number ratio calculating circuit 26 is provided, the threshold value is selected by inputting the parameter value 25.
Even if the target area does not exist in the measurement image 7, erroneous detection of the background area is prevented.

[0056]

According to the first aspect of the present invention, even when the target area is smaller than the background area, the high-frequency components of the edges of the target area and the background area appear in the significant pixel number ratio. By doing so, only the target area can be stably extracted.

According to the second aspect, the extreme value of the ratio of the number of significant pixels due to the high-frequency component of the edge of the background area can be removed by the parameter value, and the background area can be removed even when the target area does not exist in the measurement image. Can be prevented from being erroneously detected.

According to the third aspect, even when the luminance value of the target area is lower than the luminance value of the background area, the high-frequency components of the edges of the target area and the background area appear in the background pixel number ratio. Thus, only the target area can be stably extracted.

According to the fourth aspect, even when the background pixel number ratio is evaluated, the extreme value of the background pixel number ratio due to the high frequency component of the edge of the background area can be removed by inputting the parameter value. Even when the target area does not exist in the measurement image, erroneous detection of the background area can be prevented.

[Brief description of the drawings]

FIG. 1 is a first embodiment of an image processing apparatus according to the present invention;
FIG. 2 is a configuration diagram showing the configuration of FIG.

FIG. 2 is a diagram for explaining the operation of the significant pixel number ratio calculating circuit according to the first embodiment of the present invention;

FIG. 3 is a second embodiment of the image processing apparatus according to the present invention;
FIG. 2 is a configuration diagram showing the configuration of FIG.

FIG. 4 is a diagram illustrating operations of a significant pixel number ratio calculation circuit and a second threshold value selection circuit according to the second embodiment of the present invention.

FIG. 5 is a third embodiment of the image processing apparatus according to the present invention;
FIG. 2 is a configuration diagram showing the configuration of FIG.

FIG. 6 is a diagram illustrating an operation of a background pixel number ratio calculation circuit according to Embodiment 3 of the present invention.

FIG. 7 is a fourth embodiment of the image processing apparatus according to the present invention;
FIG. 2 is a configuration diagram showing the configuration of FIG.

FIG. 8 is a diagram illustrating operations of a background pixel number ratio calculation circuit and a second threshold value selection circuit according to Embodiment 4 of the present invention.

FIG. 9 is a diagram illustrating an operation mode of a conventional image processing apparatus.

FIG. 10 is a diagram illustrating a configuration of a conventional image processing apparatus.

FIG. 11 is a diagram showing an example of a grayscale image by a conventional image processing apparatus.

FIG. 12 is a diagram illustrating an example of a histogram measurement distribution of a measurement image by a conventional image processing apparatus.

[Explanation of symbols]

REFERENCE SIGNS LIST 1 aircraft, 2 image processing device, 3 target, 4 infrared imaging circuit, 6 gate setting circuit, 8 histogram measurement circuit, 9 threshold value calculation circuit, 11 binarization circuit, 13 center of gravity value calculation circuit, 20 low-pass filter circuit, 21 Significant pixel number ratio calculation circuit, 23 first threshold value selection circuit, 24 second threshold value selection circuit, 26 background pixel number ratio calculation circuit.

Claims (4)

[Claims] 1. An infrared imaging means for outputting a voltage value that varies according to the intensity of infrared radiation as a grayscale image for each pixel, and a measurement area is set in the grayscale image output from the infrared imaging means. Measurement area setting means for outputting the measurement frequency, first histogram measurement means for measuring the frequency of occurrence for each luminance value of the measurement image, and low-frequency component for extracting the low-frequency component of the measurement image and outputting it as a low-frequency component image A component image extracting unit, a second histogram measuring unit that measures the frequency of occurrence for each luminance value of the low-frequency component image, and a measurement value of the first histogram measuring unit that is input and the measured image is divided for each luminance value. The number of significant pixels in the case of binarization and the measured value of the second histogram measuring means are input, and the number of significant pixels in each luminance value is calculated from the number of significant pixels in the case where the measured image is binarized for each luminance value. Calculate ratio Means for calculating a significant pixel number ratio, and calculating a luminance value at which the significant pixel number ratio becomes an extreme value.
An image processing apparatus comprising: a threshold value selecting unit that selects a luminance value close to the threshold value as a threshold value; and a binarizing unit that compares the luminance value of the measurement image with the threshold value and outputs a binary image. . 2. An infrared imaging means for outputting a voltage value that changes according to the intensity of infrared radiation as a gray image for each pixel, and a measurement area is set in the gray image output from the infrared imaging means. Measurement area setting means for outputting the measurement frequency, first histogram measurement means for measuring the frequency of occurrence for each luminance value of the measurement image, and low-frequency component for extracting the low-frequency component of the measurement image and outputting it as a low-frequency component image A component image extracting means, a second histogram measuring means for measuring the frequency of occurrence for each luminance value of the low-frequency component image, and a measured value of the first histogram measuring means, and the measured image is divided by the luminance value. The number of significant pixels when binarized and the measured value of the second histogram measuring means are input, and the number of significant pixels when the measured image is binarized for each luminance value is converted to a significant pixel for each luminance value. Calculate number ratio Means for calculating the ratio of significant pixels to be output,
Comparing a parameter value with an extreme value of the ratio of significant pixels, and selecting a brightness value of an extreme value that does not exceed the parameter value and is close to 1 as a threshold value, and a brightness value of the measurement image and the threshold value. And a binarizing unit that outputs a binary image by comparing the two. 3. An infrared imaging means for outputting a voltage value that changes according to the radiation intensity of infrared rays as a grayscale image for each pixel, and a measurement area is set in the grayscale image output from the infrared imaging means. Area setting means for outputting as an image, and a first means for measuring an occurrence frequency for each luminance value of the measurement image.
A low-frequency component image extracting means for extracting a low-frequency component of the measurement image and outputting the low-frequency component image as a low-frequency component image, and a second histogram for measuring an occurrence frequency for each luminance value of the low-frequency component image Measuring means, inputting the measured value of the first histogram measuring means, inputting the number of background pixels when the measured image is binarized for each luminance value, and inputting the measured value of the second histogram measuring means; Background pixel number ratio calculating means for calculating a background pixel number ratio for each luminance value from the number of background pixels when the measurement image is binarized for each luminance value, and a luminance value at which the background pixel number ratio is an extreme value And a binarizing unit that outputs a pixel whose luminance value of the measurement image is smaller than the threshold value as a binary image. An image processing apparatus characterized by the above-mentioned. 4. An infrared imaging means for outputting a voltage value changing according to an infrared radiation intensity as a grayscale image for each pixel, and a measurement area is set in the grayscale image output from the infrared imaging means. Area setting means for outputting as an image, and a first means for measuring an occurrence frequency for each luminance value of the measurement image.
A low-frequency component image extracting means for extracting a low-frequency component of the measurement image and outputting the low-frequency component image as a low-frequency component image, and a second histogram for measuring an occurrence frequency for each luminance value of the low-frequency component image Measuring means, inputting the measured value of the first histogram measuring means, inputting the number of background pixels when the measured image is binarized for each luminance value, and inputting the measured value of the second histogram measuring means; Background pixel number ratio calculating means for calculating a background pixel ratio for each luminance value from the number of background pixels when the measurement image is binarized for each luminance value, and comparing an extreme value of the parameter value and the background pixel ratio. Threshold value selecting means for selecting, as a threshold, an extreme luminance value which does not exceed the parameter value and is close to 1, and binarization for outputting a pixel whose luminance value of the measurement image is smaller than the threshold value as a binary image Having means The image processing apparatus according to claim 3, wherein.