Self-adapting image enhancement method based on related coefficient

Abstract

The invention discloses a self-adapted image enhancement method based on a related coefficient. In the method, firstly, three low pass filter processes are performed on an original image to obtain three layers of low frequency image, and three layers of high frequency image are obtained by performing vector subtraction operation on adjacent layers of the low frequency image; secondly, relative coefficient of each layer is calculated by performing division on points of adjacent layers of the high frequency image, details and noises are separated and enhanced or inhibited via the related coefficient, and vector addition operations are performed on processed high frequency image of each layer and the third low frequency image to obtain detail-sharpened such as edge-sharpened image; and finally, visual behave optimized theory is applied on the method to perform contrast self-adapted enhanced process on the sharpened image so as to obtain enhanced desired image. The invention can separate noises and details and can self-adapt process images on various weather conditions, and the enhanced image has good texture, edge details and desired contrast.

Description

technical field [0001] The invention relates to an adaptive image enhancement method, in particular to a method for sharpening an image by using the correlation coefficient between images of various layers, and applying the theory of optimal visual performance to adaptively adjust the contrast of the sharpened image, which belongs to digital image Dealing with technical fields. Background technique [0002] Image enhancement refers to the use of corresponding technical means to emphasize certain features in the image and suppress other information according to the specific needs of human eye observation and / or machine further analysis and processing, so as to improve contrast and sharpen edge details. There are many existing image enhancement methods, such as unsharp mask method, histogram equalization method, Lee image enhancement algorithm, genetic algorithm and so on. However, most of these existing methods face two basic problems: 1. When the gray value of the enhanced ...

AI Technical Summary

Problems solved by technology

However, most of these existing methods face two basic problems: 1. When the gray value of the enhanced image exceeds the gray scale range, the image will be renormalized, resulting in a decrease in contrast and loss of information, such as anti-sharpness in the spatial domain. 2. Details and noise are difficult to distinguish, resulting in conflicts between enhancing details and suppressing noise, such as histogram equalization in the airspace, etc.

[0004] However, the above methods still cannot accurately distinguish between details and noise

This is because there are a lot of edge, texture and noise information in the low-level high-frequency image at the same time. The above method suppresses all the information in the low-level as noise, which will cause the edge and texture information distributed in the low-level to be lost.

Moreover, this method requires the user to input parameters according to the exposure of the image to be processed to achieve contrast adjustment, which usually requires multiple experiments to obtain an enhanced image with good visual effect, which requires a high level of personal experience for the user, so there are also Inconvenient to use

Images