Filtering method and filtering device of G points in RAW image

Abstract

The embodiment of the invention discloses a filtering method of G points in RAW images and also discloses a filtering device of the G points in the RAW images, wherein the method comprises the following steps of establishing an M*N window by using the current G point as a center, and calculating the absolute gradient values of horizontal, vertical, inclined 45-degree and inclined 135-degree directions according to the distributing situation of R (Red), G (Green) and B (Blue) color channels in the M*N window, wherein the horizontal and vertical directions are used as a first direction group, the inclined 45-degree and inclined 135-degree directions are used as a second direction group, and both the M and the N are positive integers; and determining which direction group is selected by adopting a hard-decision method according to the absolute gradient values of the four directions. By adopting the invention, the proper direction group can be selected by accurately identifying which group's detail is in the ascendance through hard decision; and in addition, the weighted mean value of the G channel is calculated through soft decision by adopting a smaller-granularity absolute-gradient-value calculating method and a G-channel mean-value calculating method, and the high-frequency detail is better reserved after filtering.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a G-point filtering method and filtering device in a RAW image. Background technique [0002] During the processing of RAW (RAW is the name of an image file format, the data of the image has not been processed), the selection of the direction is involved. [0003] In prior art 1, when selecting between different directions, a direction is selected according to the minimum absolute value of the direction gradient, and then the pixel (pixel) value in this direction is calculated. When the pixel sample value in the selected direction is small, the absolute value of the gradient will be small, which cannot be identified in the prior art. [0004] In the second prior art, when selecting between different directions, the selection is made according to the smallest absolute value of the gradient of the direction / average pixel value of the direction, and then the average pixel value in t...

AI Technical Summary

Problems solved by technology

In the RAW image, the calculation of the mean value of the pixel in each direction is not consistent, and when the length of the direction is short, the calculated mean value fluctuates greatly, which ultimately affects the accuracy of the absolute value of the gradient relative to the minimum method.

Images