Realization method and device of heavy mosaics

Abstract

The invention discloses a method and a device for realizing heavy mosaics. The method comprises the following steps of: calculating gradients of a unit where a pixel point needing green interpolation in a multi-unit CFA image is located in multiple directions, and calculating texture gradient matrixes of the unit and adjacent units; calculating a texture similarity coefficient according to the texture gradient matrixes of the unit and the adjacent unit, determining an interpolation direction according to the texture similarity coefficient and the gradient in each direction, and performing green component interpolation on the pixel point of the unit; respectively carrying out blue component interpolation and red component interpolation on the pixel points at the red and blue component positions according to the full-resolution green channel obtained by interpolation, and carrying out red and blue component interpolation on the pixel points at the green component position according to the blue and red channels obtained by interpolation; and combining all color channels to obtain a re-mosaic image. Therefore, according to the method, the algorithm complexity can be reduced, the hardware overhead can be reduced, hardware realization is facilitated, and the processing flow has real-time performance.

Description

technical field [0001] The present invention relates to the technical field of image processing, and in particular, to a method and device for realizing re-mosaic. Background technique [0002] In order to reduce the cost, reduce the volume and facilitate hardware processing, the raw (Raw) image collected by a single image sensor (Sensor) is called a Color Filter Array (CFA) image, and the CFA image is a mosaic arrangement image. The color filter array covered in front of the image sensor filters the captured light according to wavelength. This method can regularly collect only one color component of multiple color components on each pixel of the image, and a specific CFA Type has a corresponding arrangement of color components. Next, the Image Signal Processor (ISP) will process the CFA image captured by the single image sensor, and obtain a complete full-color image after demosaicing operation, in which the missing color of each pixel in the CFA image is obtained by The ...

AI Technical Summary

Problems solved by technology

It has been found in practice that in order to make the demosaic module of the image signal processor of the traditional single-cell color filter array (one-cell CFA) compatible to process images collected by multi-cell CFA (that is, multi-cell CFA images), in the existing implementation Using eight directions for interpolation of the green channel, this method has the problems of high algorithm complexity and high hardware overhead, or, in the existing implementation method, the multi-unit CFA image is split into multiple low-resolution Bayer images with the same arrangement mode format CFA images, these low-resolution Bayer-format CFA images are demosaiced and turned into RGB images, and finally fused into a full-resolution RGB image. This method has the problems of high algorithm complexity and high hardware overhead.

It can be seen that the current way of image signal processors to process multi-unit CFA images has the problems of high algorithm complexity and high hardware overhead, which is not conducive to the realization of hardware

Images