ISP color interpolation method and device based on self-adaptive Gaussian kernel

Abstract

The invention discloses an ISP color interpolation method and device based on an adaptive Gaussian kernel, and the method comprises the steps: S1, obtaining the original data of a to-be-processed image, and respectively obtaining R, G and B channel pixels after RGB channel separation; s2, traversing each to-be-interpolated pixel point on the G channel, respectively calculating correlation information between each to-be-interpolated pixel point and a neighborhood, and generating an adaptive Gaussian kernel; s3, performing interpolation on each pixel point to be interpolated according to the generated adaptive Gaussian kernel to obtain an interpolated G channel; and S4, respectively interpolating the R channel and the B channel to obtain the interpolated R channel and B channel, and fusing the R channel and B channel to obtain the interpolated RGB data. The method and device can sense the edge region and the non-edge region in the image and reserve the edge information of the image, aresimple in implementation method, are low in calculation complexity, are high in execution efficiency, and are good in color fidelity effect.

Description

technical field [0001] The present invention relates to the technical field of ISP (Image Signal Processing, image signal processing), in particular to an ISP color interpolation method and device based on an adaptive Gaussian kernel. Background technique [0002] In digital cameras, a single CCD or CMOS image sensor is usually used to achieve image acquisition, but the image sensor can only acquire one component of RGB color on one pixel, and three image sensors are required to acquire different color components. Considering the implementation cost And the degree of complexity, usually a layer of color filter array CFA is covered on the surface of the sensor as a filter layer, the filter points of the filter layer correspond to the pixels of the image sensor one by one, and each filter point can only pass through red, green, blue One of the three kinds of light, that is, the filtering layer can obtain a single color component at each pixel coordinate, and estimate the other...

AI Technical Summary

Problems solved by technology

This type of bilinear interpolation method can be simplified as a fixed template convolution, which is convenient for parallel implementation and can meet the requirements of low calculation and high frame rate. There will be obvious zipper effect (zipper effect) and false color phenomenon (false color) at the edge

Although the edge aliasing effect can be identified and weakened by post-processing, this type of post-processing, on the one hand, destroys the proper process of the ISP, increases the delay link in the pipeline processing of image signals, and requires a large number of algorithms parameters, which further increases the workload and difficulty of parameter setting; on the other hand, the distortion caused by the interpolation algorithm is handed over to the post-processing to solve, which is equivalent to destroying the image structure first and then rebuilding it, although it can correct the aliasing to a certain extent, etc. Distortion phenomenon, but it is easier to cause damage to the original structure of the image

Images