JP2013058969A - Image processing apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a motion vector reliably even in a scene where there is only color difference with no difference of luminance between patterns.SOLUTION: The color component difference from the maximum value to the minimum value of one pixel in input video signals R, G, B is extracted as a color contrast by a color contrast extraction unit 11. The contrast extracted by the color contrast extraction unit 11 is doubled, for example, by a contrast ratio enhancement unit 12. The luminance signal components of the video signals R, G, B are generated by a luminance signal generation unit 14. The contrast output from the contrast ratio enhancement unit 12 and the luminance signal generated by the luminance signal generation unit 14 are added by an addition unit 13, the addition value is supplied to a below gray scale size adjustment unit 15, and adjusted to go below the data size of gray scale. Consequently, a signal for motion detection in a scene where there is only color difference with no difference of luminance between patterns can be obtained.

Description

  Embodiments described herein relate generally to an image processing apparatus that processes an image in which a motion vector can be detected from an image having no luminance difference between patterns.

  Conventionally, as a means for calculating an image motion vector, a block matching method and a gradient method for detecting a motion vector for each pixel or small block between two consecutive frames are known. When obtaining a motion vector of a color image, it is originally desirable to obtain a motion vector from each of three signals such as the three primary color signals R, G, and B. However, in order to simplify the processing, means for detecting a motion vector by detecting one system of motion vectors has been proposed.

  In one system of motion vector detection, a pixel block is defined, the threshold value of the luminance gradient component of the luminance signal of the pixel in the block is determined, and when it is below the threshold value, the color saturation signal of the color difference signal is added to the luminance signal in the block, A motion vector is detected using this luminance signal. This motion vector detection requires a large input dynamic range, and the cost is higher than that for detecting a motion vector from a normal luminance signal. In addition, if the threshold determination for determining whether or not a color saturation signal is added malfunctions due to noise or the like, a malfunction occurs in motion vector calculation, and the threshold determination malfunction becomes vulnerable.

JP 2005-354370 A

  Since the conventional motion vector detection technology defines pixel blocks and is simulated in units of blocks, if the level changes due to noise or the like, a luminance difference occurs in the block, and an accurate motion vector cannot be detected. There was a problem.

  This embodiment is to provide an image processing apparatus capable of obtaining an image capable of reliably detecting a motion vector even in an image of a scene having only a color difference having no luminance difference between pictures.

  According to the embodiment, the color contrast extraction unit that extracts the color component difference between the maximum value and the minimum value of one pixel in the input video signal as the color contrast, and the contrast extracted by the color contrast extraction unit is a constant. A contrast ratio improving unit for improving the luminance signal, a luminance signal generating unit for generating a luminance signal component of the video signal, an improved contrast output from the contrast ratio improving unit, and a luminance signal generated by the luminance signal generating unit. An adding unit for adding, and a grayscale size or smaller adjusting unit that outputs the adjusted value so that an added value output from the adding unit is equal to or smaller than a grayscale data size.

It is a block diagram for demonstrating 1st Embodiment regarding an image processing apparatus. It is a flowchart for demonstrating the conversion process of 1st Embodiment. It is a block diagram for demonstrating 2nd Embodiment regarding an image processing apparatus. It is a flowchart for demonstrating the conversion process of 2nd Embodiment. It is explanatory drawing for demonstrating an example of the ground color cancellation of 2nd Embodiment. It is explanatory drawing for demonstrating the other example of the ground color cancellation of 2nd Embodiment.

  Hereinafter, embodiments will be described in detail with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram for explaining a first embodiment relating to an image processing apparatus.

  The image processing apparatus shown in FIG. 1 supplies, for example, video signals R, G, and B of three primary colors captured from a video camera to the color contrast extraction unit 11, and extracts the contrast of the R, G, and B signals. The color contrast extraction unit 11 extracts color contrast from the pixel data Ri, Gi, Bi based on the R, G, B signals. The extracted color contrast is improved in color contrast via the color contrast ratio improving unit 12 and supplied to the adding unit 13. The color signal may be a video signal composed of a luminance signal and a color difference signal.

  Here, the color contrast refers to “color vividness”, and the difference (max (r, g, b) −min) between the maximum value and the minimum value component in the R, G, B signals of a certain pixel. (r, g, b)) will be referred to as color contrast. Therefore, when R, G, and B are 24-bit color, 255 is the maximum color contrast and the maximum vividness.

  Further, the input video signals R, G, and B generate luminance signals from the pixel data Ri, Gi, and Bi based on the R, G, and B signals via the luminance signal generation unit 14, and add them to the addition unit 13. Supply. The adding unit 13 adds the contrast signal for improving the contrast ratio obtained through the color contrast ratio improving unit 12 and the luminance signal generated through the luminance signal generating unit 14.

  The added color contrast signal and luminance signal are supplied to the grayscale size or smaller adjustment unit 15. The grayscale size or less adjustment unit 15 adjusts the value of the addition signal output from the addition unit 13 to a predetermined bit width (grayscale size) or less, for example, 255 or less in the case of 8 bits. (= Brightness value) This process is performed for all pixels in units of pixels. The output of the adjustment unit 15 below the grayscale size is output as a motion detection signal to a motion vector detection unit (not shown) in the next stage.

  Next, the details of the processing in each block of FIG.

  First, in the process S1, the color contrast is extracted by obtaining the maximum value-minimum value in the pixel data Ri, Gi, Bi based on the color signal data R, G, B supplied to the color contrast extraction unit 11. Do. In the process S2, the extracted color contrast signal is increased by accumulating the color contrast ratio, for example, twice by the color contrast ratio improving unit 12.

  In the process S3, the luminance signal generation unit 14 to which the R, G, and B signals are supplied generates a luminance signal by obtaining the sum (R + G + B) of the supplied R, G, and B. Although the above sum is used to simplify the calculation, y = 0.3 · R + 0.6 · G + 0.1 · B, which is a luminance value calculation formula in the NTSC system, may be used.

  Next, in the process S4, the color contrast signal obtained in the process S2 and the luminance signal obtained in the process S3 are added by the adding unit 13, and the addition result is supplied to the grayscale size or less adjusting unit 15 in the next stage.

In the next processing step S5, the grayscale size or smaller adjustment unit 15 adjusts the grayscale size to be equal to or smaller than the grayscale size because the added value in the processing S4 may exceed the grayscale size. Examples of the adjustment method are (1) to (3).
(1) Limit by gray scale size. For example, if it exceeds 255, a limit of 255 is applied.
(2) Divide by a certain value and then limit by gray scale size. For example, when dividing by 2 and still exceeding 255, a limit of 255 is applied.
(3) Divide by a value that does not exceed the grayscale size. For example, when the constant of the color contrast ratio improving unit 12 is 2 or less, dividing by 4 does not exceed 255, and no limit is required.

  Here, (1) of the adjustment method will be described with reference to FIG. Assume that the luminance values of the color signals R, G, and B are (100, 50, 150), respectively. In this case, since the maximum luminance value is 150 for the B component and the minimum luminance value is 50 for the G component, the color contrast is 150−50 = 100, and this value is output from the color contrast extraction unit 11.

  The color contrast value 100 output from the color contrast extraction unit 11 is improved to the double (2 × 100 = 200) contrast given as a constant example by the color contrast ratio improvement unit 12.

  The adder 13 adds the luminance values (100 + 50 + 150) of the color signals R, G, and B generated from the luminance signal generator 14 and the color contrast value 200 output from the color contrast extractor 11 (100 + 50 + 150 + 200). Output the value. Therefore, the addition value 500 of the addition unit 13 has a relationship of 500> 255, which is higher than the maximum value 255 of the gray scale size. Therefore, the portion exceeding the gray scale size 255 is limited by the adjustment unit 15 below the gray scale size, and an output of 255 is obtained from the output.

  The above processes S1 to S4 are performed for all the pixels, and the motion vector detection by one gray system is finished.

  As described above, the motion vector detection unit that obtains the motion vector based on the luminance signal in the next stage of the grayscale size or less adjustment unit 15 by adding the luminance difference to the same luminance by the color contrast strength is used in the motion vector detection unit. An image effective for vector detection can be obtained.

  In addition, when noise is generated by converting the luminance in units of one pixel in the image block, there is a defect in the luminance value when viewed from the one pixel, but when the entire block is viewed, a luminance value close to the original value is obtained. And the influence on noise can be suppressed.

  By the way, in each of the adjustment methods (2) and (3), a value not exceeding 255 is output from the adjustment unit 15 below the gray scale size, and a luminance signal that can easily obtain a motion vector can be obtained. .

  In this embodiment, since grayscale conversion (luminance value) is obtained for each pixel, even when noise is mixed in a certain pixel and an abnormal luminance value is obtained, a normal luminance value is obtained for other pixels. Can do. Therefore, even if noise is mixed during motion vector extraction by the gradient method, there is a possibility that a normal motion vector can be extracted because there is a normal value pixel in a predetermined area, and the influence on noise can be reduced. It becomes.

(Second Embodiment)
FIG. 3 is a block diagram for explaining a second embodiment related to the image processing apparatus. In this embodiment, the functions are further expanded from those of the first embodiment. The same reference numerals are given to the same components as those in the first embodiment, and the description here will focus on different parts. I will explain.

  In this embodiment, an image taken from a video camera supplied to the color contrast extraction unit 11 and the luminance signal generation unit 14 is processed in consideration of a color other than so-called gray where the ground area of the road is an asphalt color. It corresponds to. That is, it has a ground color canceling function that lifts an object from the ground surface of a color having a high color contrast as a whole.

  Processing in this embodiment will be described with reference to FIG. That is, in the processing S0, the ground color canceling unit 31 to which the three primary color video signals R, G, B captured from the video camera, for example, are input extracts the ground colors Rg, Gg, Bg. The extracted ground colors Rg, Gg, Bg are stored in a temporarily storable memory such as a flash memory. The extraction of the ground color is performed using the most common color among the ground colors Rg, Gg, and Bg or the average value of the ground colors. Therefore, the input pixel data Ri, Gi, Bi is canceled based on the ground colors Rg, Gg, Bg stored in the memory. From the output of the ground color cancellation unit 31, output pixel data Ro, Bo, Go from which the ground colors Rg, Gg, Bg are canceled is output.

  In the subsequent processes S1 to S4, the output pixel data Ro, Bo, Go of the ground color canceling unit 31 is obtained as color contrast signals and luminance signals as input signals at the time of color contrast extraction and luminance signal generation. Processing similar to that of the embodiment is performed, and a value adjusted to a gray scale size of 255 or less is output, so that a luminance signal that can easily obtain a motion vector can be obtained.

  Next, a more specific example of ground color cancellation when the ground color Rg, Gg, Bg is included in the video signals R, G, B input to the ground color cancellation unit 31 will be described with reference to FIG. .

  This ground color canceling example is extracted as the largest component Cg_max in the ground colors Rg, Gg, Bg in the ground area in the video signals R, G, B, and the ratio of the component colors of the ground area extracted here Thus, the output pixel data Ro, Bo, and Go are respectively calculated as follows.

Ro = Ri · Cg_max / Rg
Go = Gi · Cg_max / Gg
Bo = Bi · Cg_max / Bg
For example, in the case of a bus lane of wine color R = 102, G = 51, and B = 51, an object having the same component is buried and cannot be detected. However, the ground color canceling unit 31 can detect the object by changing the RGB component of the object as follows.

That is, when the component of the object is R = 102, G = 51, B = 102,
Ro = 102 · 102/102 = 102
Go = 51 · 102/51 = 102
Bo = 102 · 102/51 = 204
Ground luminance (Yg) = 102 + 51 + 51 = 204
Object luminance (Yo) = 102 + 102 + 204 = 408
It becomes.

  Since the object brightness Yo is twice the value of the ground brightness Yg as described above, it is possible to improve the detection of an object that emphasizes the difference between an object having a high color contrast and a color contrast having a high color contrast. be able to.

  FIG. 6 is an explanatory diagram for explaining another more specific example of the ground color cancellation.

  This ground color cancellation example is extracted as the largest component Cg_max in the ground colors Rg, Gg, Bg in the ground region in the video signals R, G, B, and is the largest component of the ground basin extracted here. The output pixel data Ro, Bo, Go are calculated by the following formulas based on the difference of Cg_max.

Ro = Ri + Cg_max−Rg
Go = Gi + Cg_max−Gg
Bo = Bi + Cg_max−Bg
For example, the luminance signal values of the ground and the object before the video signals R, G, and B are converted through the ground color canceling unit 31 are respectively
Ground luminance (Yg) = 102 + 51 + 51 = 204
Object luminance (Yo) = 102 + 51 + 102 = 255
And When the output pixel data Ro, Bo, Go is calculated by the difference of the largest component Cg_max of the extracted ground basin,
Ro = 102 + 102−102 = 102
Go = 51 + 102−51 = 102
Bo = 102 + 102−51 = 153
Ground luminance (Yg) = 102 + 51 + 51 = 204
Object luminance (Yo) = 102 + 102 + 153 = 357
It becomes.

  Thus, the object brightness Yo can be enhanced from 255 to 357 with respect to the ground brightness Yg. For this reason, it is possible to improve the detection of an emphasized object that is different from an object having a high color contrast in a different color, even in a picture having a high color contrast as a whole.

  In each of the above examples, the largest component Cg_max in the ground colors Rg, Gg, and Bg in the ground region is extracted. However, based on the average color of the ground colors Rg, Gg, and Bg, the video signal is expressed in proportion or difference. Even if it is calculated, it is possible to enhance the object brightness.

  In this embodiment, in addition to the same effects as those of the first embodiment, an object having the same color can be detected on the colored ground. Further, even on a colored ground, an object different from the ground can be detected.

  Although several embodiments have been described, these embodiments have been presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, combinations, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 11 Color contrast extraction part 12 Color contrast ratio improvement part 13 Addition part 14 Luminance signal generation part 15 Gray scale size or less adjustment part 31 Ground color cancellation part

Claims (7)

A color contrast extraction unit that extracts a color component difference between the maximum value and the minimum value of one pixel in the input video signal as a color contrast;
A contrast ratio improvement unit for improving the contrast extracted by the color contrast extraction unit by a constant amount;
A luminance signal generator for generating a luminance signal component of the video signal;
An adding unit that adds the improved contrast output from the contrast ratio improving unit and the luminance signal generated by the luminance signal generating unit;
An image processing apparatus comprising: a grayscale size or less adjustment unit that outputs an adjustment value adjusted so that an addition value output from the addition unit is equal to or less than a grayscale data size.   2. The image processing apparatus according to claim 1, wherein a ground color canceling unit that raises an object from a ground having a color with a high color contrast in the video signal is arranged in a preceding stage of the color contrast extracting unit.   The image processing apparatus according to claim 2, wherein the ground color canceling unit extracts an average color or a most frequent component color in the ground area in the video signal, and calculates a video signal at a ratio of the component color.   The image processing apparatus according to claim 2, wherein the ground color cancellation unit extracts a component color of an average color or a largest number of colors in the ground region in the video signal, and calculates a difference between the component color and the video signal. .   The image processing apparatus according to any one of claims 1 to 4, wherein the grayscale size or smaller adjustment unit is adjusted to a size limited by the grayscale size.   The image processing apparatus according to claim 1, wherein the grayscale size or smaller adjustment unit adjusts to a size limited by the grayscale size after dividing by a certain value.   The image processing apparatus according to any one of claims 1 to 4, wherein the grayscale size or smaller adjustment unit adjusts the size divided by a value not exceeding the grayscale size.

Images