CN103780890A - White balance adjustment method - Google Patents

Abstract

A method for adjusting white balance, extracting the color scale values of the RGB channels in the Raw image, obtaining the maximum value, minimum value and average value of the color scale values of each channel, and then defining the threshold modulation coefficient of the corresponding channel by the above three values, Redefine the maximum value threshold for reference white and the minimum value threshold for reference black of the corresponding channel according to the respective modulation coefficients of the RGB channels, and finally complete the white balance adjustment of the image according to the reference white and reference black.

Description

White Balance Adjustment Method

technical field

The present invention relates to an image processing technology, in particular to a white balance adjustment method in the image processing technology.

Background technique

Generally speaking, the color of light reflected by an object depends on the color of the light source. The human brain can detect and correct such color changes, so the human eye can adapt to such color changes whether in sunlight, cloudy weather, indoor incandescent or fluorescent lighting, without affecting the whiteness that people see object. However, as far as the digital image capture device is concerned, the "white" produced by different light sources is still not the same in color. For example, pictures taken under tungsten (bulb) lighting may have a yellowish cast, while pictures taken under other conditions may have a blue or red cast.

In order to be close to human vision, the digital image acquisition device must imitate the human brain and adjust the color according to the light, and define what is white, so that the white seen by the naked eye can be presented in the final photo. This adjustment is called " White Balance".

In the prior art, it is proposed to use the histogram stretching method (Color Histogram Stretching, CHS) to adjust the image white balance, and requantize the histogram of the RGB three-channel (Red, Green, Blue) of the image without white balance to 0- The range of 255, the adjustment results show that this method has small error and real-time characteristics.

On the histogram, the direction of the larger color gradation value and the direction of the smaller color gradation value define two ratios of the same size as the threshold, and the average value of the color gradation values corresponding to the pixels within these two thresholds is the reference white and the reference black respectively. , to remap the gradation value according to the reference white and reference black to complete the automatic white balance.

On the basis of these methodologies, although reference white and reference black can be found, since the thresholds for reference white and reference black to be obtained are fixed, the obtained reference white and reference black are easily restricted and fixed within a certain range Inside. On the other hand, if the histogram is normally distributed, the obtained reference white and reference black are relatively stable and not easily affected by noise; but if the main peak is distributed to the left, the obtained reference white is too large; otherwise , the main peak is distributed to the right, and the reference black that is too small is obtained. In the result, the contrast after white balance adjustment is poor.

Contents of the invention

In view of this, it is necessary to provide a white balance adjustment method with a variable threshold.

A method for adjusting white balance, comprising the steps of: setting the initial maximum threshold Rmax and the minimum threshold Rmin of R, G, and B channels in a Raw image, and Rmax=Rmin; extracting R, G, and B in the Raw image The color scale value of the channel, and obtain the minimum value a, maximum value b, and average value c of the color scale value of the corresponding channel respectively; respectively define the modulation coefficient r of the R, G, and B channels, r=(c-a)/(b-a ); redefine the new maximum threshold Rmax' and minimum threshold Rmin' of the R, G, and B channels respectively, if r<0.5, then Rmin'=r×Rmin and Rmax'=(2.0-r)×Rmax, and vice versa , Rmin'=(2.0-r)×Rmin and Rmax'=r×Rmax; obtain the average of the color scale values corresponding to the pixels in the new maximum threshold Rmax' and minimum threshold Rmin' of the R, G, and B channels respectively value and then obtain the reference white and reference black of the corresponding channel, and use the histogram extension method to complete the white balance adjustment of the corresponding channel according to the reference black and reference white of the R, G, and B channels.

Compared with the prior art, in the white balance adjustment method of the embodiment of the present invention, the maximum value threshold and the minimum value threshold of the RGB channel are set according to the maximum value, minimum value and average value of the color scale value of the corresponding channel, Different channels have different maximum thresholds and minimum thresholds according to their respective color scale values, and then adjust the white balance of the corresponding channel according to the average value of the color scale values corresponding to the adjusted maximum value threshold and minimum value threshold. Therefore, the white balance adjustment method can better reflect the real situation of the image, and can improve the contrast of the image after the white balance adjustment.

Description of drawings

Fig. 1 is a flow chart of implementing the white balance adjustment method of the present invention.

Detailed ways

Please refer to FIG. 1, the white balance adjustment method provided by the embodiment of the present invention is used to adjust the white balance of the image, and the white balance adjustment method includes the following steps:

S1, load the Raw image, and set the initial maximum threshold Rmax and minimum threshold Rmin of the R, G, and B channels respectively, and Rmax=Rmin, the average value of the color scale value within the maximum threshold represents the reference white, and the minimum The average value of the color scale values within the value threshold represents the reference black.

S2. Extract the color scale values of the R, G, and B channels in the Raw image, and obtain the minimum value a, maximum value b, and average value c of the color scale values of the R, G, and B channels, respectively.

S3, respectively defining the modulation coefficient r of the R, G, and B channels, r=(c-a)/(b-a).

S4, redefine the new maximum threshold Rmax' and minimum threshold Rmin' of the R, G, and B channels respectively, if r<0.5, then Rmin'=r×Rmin and Rmax'=(2.0-r)×Rmax, and vice versa , Rmin'=(2.0-r)×Rmin and Rmax'=r×Rmax.

S5, obtain the average value of the color scale values in the new maximum threshold Rmax' and minimum threshold Rmin' of the R, G, and B channels respectively, and then obtain the reference white and reference black of the corresponding channels, according to the R, G, and B channels Use the histogram extension method to complete the white balance adjustment of the corresponding channels respectively for reference black and reference white.

In step S1, the values of the initial maximum threshold Rmax and minimum threshold Rmin in the R, G, and B channels of the Raw image can be customized by the user, for example, Rmax=Rmin=0.01, Rmax=Rmin=0.02, Rmax =Rmin=0.03 and so on. In this embodiment, Rmax=Rmin=0.01.

The maximum value threshold Rmax is: capture a certain number of pixels from the maximum color scale value to the direction where the color scale value gradually decreases, and the ratio of the number of pixels to the total number of pixels in the Raw image is the maximum value threshold. The pixels within the range corresponding to the maximum value threshold Rmax refer to the above-mentioned certain number of pixels taken in a direction from the maximum color scale value to a gradually decreasing color scale value.

The minimum value threshold Rmin is: capture a certain number of pixels from the minimum color scale value to the direction where the color scale value gradually increases, and the ratio of the number of pixels to the total number of pixels in the Raw image is the minimum value threshold. The pixels within the range corresponding to the minimum value threshold Rmin refer to the predetermined number of pixels taken in a direction from the minimum color scale value to the gradually increasing color scale value.

In this embodiment, Rmax=Rmin, that is, the number of pixels in the ranges corresponding to the minimum value threshold Rmin and the maximum value threshold Rmax are the same. However, the ranges spanned by the pixel color scale values corresponding to the minimum value threshold Rmin and the maximum value threshold Rmax may be different.

In step S2, a Raw image is composed of multiple pixels (pixels), and according to the color filter covered on the surface of the image sensor, the Raw image can usually be sub-pixel R (Red, red), G (Green, green ), B (Blue, blue) raw data representation.

Specifically, first extract the color scale value of the R channel in the Raw image, compare the color scale value to obtain the maximum value, the minimum value, and calculate the average value through the maximum value and the minimum value; then refer to the method of the R channel to extract the G, The color gradation value of the B channel is compared and calculated to obtain the maximum, minimum and average value of the color gradation values of the G and B channels.

In this step, the extraction order and calculation order of the color scale values of the R, G, and B channels are not limited. For example, the color scale values of the G channel or the B channel may be extracted first to compare and calculate their maximum, minimum, and average values. In addition, the color scale values of the R, G, and B channels can also be extracted first, and then the values of the corresponding channels can be calculated.

In step S3, according to the obtained maximum value, minimum value and average value, the modulation coefficient r of the corresponding channel is obtained by the formula r=(c-a)/(b-a).

In this step, the order of calculating the corresponding modulation coefficients r of the R, G, and B channels is not limited.

In step S5, use the new maximum value threshold and minimum value threshold of each channel to calculate the average value of the color scale values in the maximum value threshold and the minimum value threshold to obtain the reference white H and reference black L of the corresponding channel, and use the reference black and reference White and use the histogram extension method to complete the white balance adjustment of the R, G, and B channels.

，其中，C表示Raw图像中每一个像素的色阶值、M为255或1023。R、G、B通道中的像素应采用相应通道的H、L，例如，计算R通道像素调整后色阶值，H、L应采用R通道的相应值。 In the histogram extension method, each pixel in the Raw image is adjusted according to the reference black and reference white, and the color scale value of the pixel after white balance adjustment is:

, where C represents the color scale value of each pixel in the Raw image, and M is 255 or 1023. The pixels in the R, G, and B channels should use the H, L of the corresponding channel. For example, to calculate the adjusted color scale value of the R channel pixel, H, L should use the corresponding value of the R channel.

It should be noted that, in steps S2-S5, the order of the white balance adjustment of the R, G, and B channels is not limited. For example, the white balance adjustment of the R channel can be completed first, and then the white balance adjustment of the G channel can be completed, and then the white balance adjustment of the G channel can be completed. White balance adjustment of the B channel; or, in accordance with the order of GRB, or the order of BGR, etc.

In the white balance adjustment method of this embodiment, the maximum value threshold and the minimum value threshold of the RGB channel are set according to the maximum value, minimum value and average value of the color scale value of the corresponding channel, and different channels are set according to their respective color scale values There will be different maximum value thresholds and minimum value thresholds, and then adjust the white balance of the corresponding channel according to the average value of the color scale values within the adjusted maximum value threshold and minimum value threshold, so this white balance adjustment method can better reflect The real situation of the image can improve the contrast of the image after white balance adjustment.

It can be understood that those skilled in the art can also make other changes within the spirit of the present invention to be used in the design of the present invention, as long as they do not deviate from the technical effects of the present invention. These changes made according to the spirit of the present invention should be included in the scope of protection of the present invention.

Claims (3)

1. A method for adjusting white balance, comprising the steps of: Set the initial maximum threshold Rmax and minimum threshold Rmin of the R, G, and B channels in the Raw image, and Rmax=Rmin; Extract the color scale values of the R, G, and B channels in the Raw image, and obtain the minimum value a, maximum value b, and average value c of the color scale values of the corresponding channels; Define the modulation coefficient r of the R, G, and B channels respectively, r=(c-a)/(b-a); Define the new maximum threshold Rmax' and minimum threshold Rmin' of the R, G, and B channels respectively. If r<0.5, then Rmin'=r×Rmin and Rmax'=(2.0-r)×Rmax, otherwise, Rmin' =(2.0-r)×Rmin and Rmax’=r×Rmax; Obtain the average value of the pixel color scale values corresponding to the new maximum threshold Rmax' and minimum threshold Rmin' of the R, G, and B channels respectively, and then obtain the reference white and reference black of the corresponding channel. According to R, G, B The reference black and reference white of the channel use the histogram extension method to complete the white balance adjustment of the corresponding channel respectively. 2 . The white balance adjustment method according to claim 1 , wherein the initial maximum value threshold Rmax and the minimum value threshold Rmin are both 0.01. 3. The white balance adjustment method according to claim 1, wherein the color scale value of the pixel after the white balance adjustment is completed by the histogram extension method is:

, where C represents the color scale value of each pixel in the Raw image, H represents the reference white, L represents the reference black, and M is 255 or 1023.

Images