Research And Implementation Of Adaptive Enhancement Algorithm For Low Quality Image

With the development of science and technology,the popularization of digital image processing technology and the promotion of imaging equipment,video image equipment is more and more used in daily life,traffic safety and medical imaging and other industries.However,in the process of image capture,due to the defects of equipment precision,the influence of light intensity and the interference of natural environment,the captured image has low brightness,image details are difficult to distinguish and more noise interference and so on.Therefore,it is necessary to enhance the image and remove the noise of the low-quality image,so as to improve the overall brightness and quality of the image to be processed.The content of this paper is to study the low quality image enhancement algorithm,mainly on the low illumination image and night image these two kinds of image brightness enhancement algorithm are studied respectively.By analyzing the defects of the existing algorithms,the brightness of the two types of images is enhanced adaptively,and the phenomenon of noise amplification is common in image enhancement,the specific research contents are as follows:Firstly,for low illumination image enhancement,a local spatial homomorphic filter algorithm with adaptive cutoff filtering is proposed in this paper.Due to the defects of spatial multi-scale homomorphic filtering algorithm,such as constrained input image size and edge block effect,a local spatial homomorphic filtering algorithm is proposed in this paper,which not only avoids the constrained input image size,and the edge block effect is also eliminated.Because the cut-off frequency point of homomorphic filter algorithm is a fixed value,it can't adapt to different characteristic images.In this paper,4/5 of the image spectrum amplitude is taken as the cutoff frequency point,so as to achieve adaptive effect.In order to improve the color saturation of the image,the S channel is filtered by local homomorphism to improve the color saturation of the image.Secondly,for the noise interference after image enhancement,a gradient domain variance guided image filtering algorithm is proposed in this paper.Since the normalized parameter ? of the guided image filtering is a fixed value,the filtering result has a halo phenomenon,so that a gradient domain guiding image filter is selected.Because the image smoothing region of gradient domain filtering image algorithm is large and the image detail information is preserved less,we propose a gradient domain variance guided image filtering algorithm.The penalty function is optimized by using the correlation of variance between pixels,so that the gradient domain variance guide image filtering can retain more image details in the filtering process.Finally,for night image enhancement,an adaptive estimation of transmittance image based on L channel is proposed in this paper.Since Dong et al.used the dark channel prior algorithm to enhance the night image,and the edge of the dark channel image of the inverted image has an expansion phenomenon,this paper proposes to replace the dark channel map with the L channel map of the inverted image.Because the L-channel map has more image detail information,and the L-channel map also saves the time required for dark channel map optimization.Due to the interference of near-white objects in the inverted image on the global atmospheric light value,the global atmospheric light estimation is optimized by combining the L channel and the mean dark channel.Because the noise is also amplified in the process of image enhancement,gradient domain variance guided image filtering is used to denoise and preserve the edge,so as to reduce the interference of noise in the image.