Image enhancement method and system based on anisotropic gradient model

Abstract

The present invention relates to an image enhancement method and system based on an anisotropic gradient model. The method includes: obtaining an image to be processed; using an L1‑L0 hybrid regularized anisotropic gradient layer decomposition model to decompose the image to be processed to obtain a preserved truth item, texture layer regularization item, and base layer regularization item; according to the fidelity item, texture layer regularization item, and base layer regularization item, the base layer information and texture layer information are determined by split Bregman iteration principle; the base layer information and texture layer information are respectively The contrast is adjusted to obtain the adjusted base layer information and the adjusted texture layer information; and the enhanced image is determined according to the adjusted base layer information and the adjusted texture layer information. Through the above method and system of the present invention, the base layer information and the texture layer information can be protected when the image layer is decomposed, and the contrast enhancement of the image can be effectively realized.

Description

technical field [0001] The invention relates to the technical field of image enhancement, in particular to an image enhancement method and system based on an anisotropic gradient model. Background technique [0002] At present, photoelectric imaging systems are widely used in many fields such as outdoor information perception, target detection and tracking, and intelligent systems. In the process of acquiring images, the imaging system inevitably encounters the situation of low ambient illumination, which makes the contrast of the target image after imaging low, which affects the later analysis and use of the image. [0003] A better means of image contrast enhancement is based on image layer decomposition, that is, the image is decomposed into base layer and texture layer, and then the two are processed separately to achieve the purpose of image information contrast enhancement. At present, the more popular layer decomposition contrast enhancement method is based on the me...

AI Technical Summary

Problems solved by technology

Although the total variational model can effectively enhance the contrast of the image and enhance the edge features of the image to a certain extent, the ladder effect of the L1 regularized total variational model will make the edge of the image base layer have a rectangular structure, such as image The sharp edge and corner features become smooth, which partially distorts the basic information of the processed image; while the L0 regularized total variation model can protect the texture layer information of the image to a certain extent, but the L0 total variation model is to uniformly control the image texture. The total number of non-zero gradients in two directions of the layer, the non-zero image gradient distribution information in a single direction cannot be known, that is, the graphic information of each direction of the texture layer cannot be protected separately, resulting in the protection of the decomposed image texture layer information insufficient

Images