Image pyramid-based image denoising method, computer device and computer readable storage medium

Abstract

The invention provides an image pyramid-based image denoising method, a computer device and a computer readable storage medium. The method comprises the following steps: acquiring an initial image, and calculating an output chromaticity value of each pixel: acquiring a hierarchical search area of a pixel to be denoised, performing downsampling on the hierarchical search area, acquiring a layer ofdownsampling image after downsampling each time, and forming a first image pyramid by multiple layers of downsampling images; performing up-sampling on each down-sampling image, subtracting a next layer of image of the down-sampling image from the up-sampled image to obtain a layer of subtracted image, and forming a second image pyramid by multiple layers of subtracted images; performing mean denoising on each layer of subtracted image of the second image pyramid and the highest layer of down-sampled image of the first image pyramid to obtain a denoised chromatic value; and calculating an output chromaticity value using the initial chromaticity value and the denoised chromaticity value. The invention further provides a computer device and a computer readable storage medium for implementingthe method. According to the method, the calculation amount of image denoising and the hardware implementation difficulty can be reduced, and the denoising quality is improved.

Description

Technical field [0001] The present invention relates to the technical field of image processing, in particular to an image denoising method based on an image pyramid, a computer device and a computer-readable storage medium for implementing the method. Background technique [0002] Many existing smart electronic devices have image capturing functions. For example, smart phones, tablet computers, driving recorders, etc. are all equipped with camera devices, and the camera devices are usually equipped with CMOS sensors to acquire images. Generally, an image includes a large number of pixels, and the color information of each pixel can be represented by RGB values ​​or YUV values. [0003] For example, the commonly used CMOS image sensors generally adopt the BAYER arrangement format. The color information of each pixel is usually the value of RGB, but the value of RGB is not the information of the three primary colors. The image will appear color distortion, so it is necessary to adju...

AI Technical Summary

Problems solved by technology

[0016] First, the denoising result of the algorithm is closely related to the size of the denoising local window. When the image noise is large, a larger local window is required to obtain a better denoising effect, but the algorithm complexity and hardware implementation cost will increase significantly

For scenes with low signal-to-noise ratio such as low illumination, the search area needs to be at least 33×33, which means that for each pixel to be denoised, more than 1088 similar image block matching operations and other related operations are required. Very large, not conducive to real-time denoising and hardware implementation

[0017] Second, when the image noise is large, the NLM algorithm can effectively remove low-frequency noise, but at the same time it will lose the high-frequency details of the image, resulting in a decrease in image clarity

This is because the search area is large at this time, and there are many similar image blocks, resulting in some false similar pixels participating in the denoising process, resulting in weakening of the image texture

[0018] Third, none of the current NLM denoising methods has a complete technical solution for RAW image denoising

Images