Adaptive Variational Correction Method for Luminance Inhomogeneity Based on Variational Framework

Abstract

The invention discloses an adaptive brightness uneven variation correction method based on the Retinex variational framework, which aims to provide a method that eliminates the uneven brightness of the image without undergoing logarithmic transformation and effectively maintains the image color and detail information. adaptive correction method. The present invention is realized through the following technologies: using an optical imaging detection system to obtain a non-uniform degradation image as an observation image; constructing an adaptive weight function according to the spatial characteristics of the observation image; using the original Retinex theory to construct a variation correction model, and adaptively controlling it through the weight function The intensity of the total variation regularization constraint of the reflection component at different pixels in the variation correction model is used, and the mean value of the reflection component is used to approximate the gray-level median constraint to prevent local overexposure; then the split Bregman iteration method and the alternating minimization method are used to solve the above variation The correction model is divided into two parts to obtain the illumination component and the reflection component, and finally the reflection component is rounded to obtain the corrected image.

Description

technical field [0001] The invention belongs to the technical field of digital image processing, and in particular relates to an adaptive brightness uneven variation correction method based on a Retinex variation framework. Among them, the Retinex variational framework means to use the variational method to solve the Retinex theory, which is commonly used in uniform light variational image processing. Background technique [0002] With the rapid development of computer technology and image processing technology, image / video-based working systems have penetrated into all walks of life, such as target recognition, automatic navigation, intelligent monitoring, terrain survey, automatic driving and civilian shooting and other fields. , has played a role that cannot be ignored in national economic and social development. However, in the imaging process of the sensor, due to the influence of environmental factors such as atmosphere and light as well as the internal factors of the...

AI Technical Summary

Problems solved by technology

Therefore, after the logarithmic transformation of the observed image, the edge and detail information of the bright area will be blurred and lost, and an important information contained in the image reflectance is the edge information in the image, so the Retinex transformation established based on the logarithmic transformation The homogenization correction image calculated by the sub-calibration model will have blurred edges and loss of details, such as the TVRE method proposed by Michael et al. and the SARV method proposed by Lan Xia et al., although better dodging correction can be obtained compared to other methods However, since the logarithmic transformation of the observed image will blur and lose the edge and detail information of the bright area, the dodging correction image also has the phenomenon of blurred edges and loss of details ( Figure 4 c,4d, Figure 5 c,5d, Image 6 c,6d, Figure 7 b,7c, Figure 8 b,8c, Figure 9 b, 9c, Figure 10 b, 10c)

Images