An image adaptive dehazing enhancement method with color constancy

Abstract

The invention provides an adaptive defogging enhancement method for an image with color constancy. On the premise of nearly not reducing an image resolution, the multispectral image with relatively low contrast, color saturation and brightness due to environmental factors such as sand dust, low illuminance, cloud fog and the like can be subjected to adaptive defogging enhancement processing, thereby achieving an image defogging effect. According to the adaptive defogging enhancement method for the image with the color constancy, a comprehensive effect of color naturalness, color richness, image contrast and brightness of the image is assessed by adopting an image fitness estimation link; adaptive solving of an image local region illuminance estimation convolution function template is realized in combination with the image fitness estimation link based on a bilateral filtering parameter set adaptive solving link of a genetic algorithm; and based on a defogging enhancement link of the image with the color constancy, a bilateral filtering convolution kernel is adopted as an image brightness channel local region illuminance estimation template, and an image brightness channel mean value is adopted as an image overall illuminance estimation value.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an image defogging enhancement method. Background technique [0002] During the flight and aerial photography of visible light cameras, under the conditions of morning and evening, clouds and fog, and uneven illumination, the multispectral images captured often show low brightness, contrast, low definition and color distortion, resulting in blurred images. The splicing process and the identification of the target by the interpreter bring certain difficulties. [0003] The Retinex image enhancement algorithm has the characteristics of high dynamic range compression and high fidelity of color. Single-scale Retinex (SSR), multi-scale Retinex (MSR), multi-scale Retinex with color restoration (MSRCR) and other methods have been developed. The Retinex algorithm can eliminate the effect of illumination changes by calculating the ratio of pixels to the weighted average of pixel...

AI Technical Summary

Problems solved by technology

However, when the algorithm is dealing with uniform scenes similar to the sky, such as snow, background walls, large areas of water, etc., since these scenes do not have obvious dark channels, the processing results are prone to image color casts

In addition, when the algorithm deals with the boundary area between the uniform background and the non-uniform scene, there will be obvious contour phenomenon. At the same time, the image processed by the dark channel prior algorithm has generally low overall brightness, and further enhancement processing is needed.

[0005] As a non-linear filter, bilateral filtering can preserve the edge and feature information of the image while filtering out the noise. It is widely used in the field of selfie filter and infrared image denoising, but it cannot enhance and denoise the image. fog treatment

[0006] Therefore, the technical problem that needs to be solved urgently is how to restore the true color information of the image and improve the clarity of the image while performing adaptive dehazing enhancement processing on the degraded image caused by dust, low illumination, cloud and fog. At the same time, it hardly reduces the resolution of the image, and can avoid the distortion of color naturalness and saturation caused by image enhancement

Images