Research On Image Sequence Enhancement Algorithms Under Non-uniform Illumination Conditions Based On Region Adaptation

With the development of mobile communication technology,image and video communication has become an indispensable part of people's daily life.With advanced acquisition equipment and professional shooting skills,people can obtain a large number of high-quality images and videos.However,not all images and videos can have satisfactory quality.Complex lighting environment or improper operation of photographers will lead to the decline of image and video quality.Such video images not only affect people's visual perception,but also increase the difficulty of subsequent processing.In view of the above situation,this paper proposes a region adaptive non-uniform illumination image / video enhancement algorithm based on the characteristics of non-uniform illumination image and video,combined with the framework of center / surround Retinex model and contrast constrained histogram equalization algorithm.The main work and innovation of this paper are as follows:(1)The traditional center surround Retinex model uses Gaussian filter to decompose the image to be enhanced into illumination component and reflection component.The illumination component estimated by this method is uniform,but in fact,the illumination in complex environment is usually non-uniform.The region adaptive image enhancement algorithm proposed in this paper estimates the illumination component through the adaptive filter,and then divides the image into blocks,sets the return illumination factor according to the statistical characteristics of each sub block,returns part of the illumination to the reflection component according to the nature of the region,and obtains the detail component of the residual part of the illumination.In this way,the estimated illumination component is closer to the real illumination of the shooting scene,and the calculated detail component can restore the essential attributes of the object to a greater extent.The detail component is compressed in a small dynamic range.In this paper,the contrast of the detail component is stretched by an improved contrast limited histogram equalization combined with a gamma correction of the mapping function.The final enhanced image is guaranteed to improve the brightness and contrast of the image without noise amplification in the dark area and over enhancement in the bright area,so that the enhanced image has better visual perception effect.(2)On the basis of the above image enhancement,this paper proposes a region adaptive enhancement algorithm for non-uniform illumination.The algorithm is divided into two parts: intra frame brightness adjustment and inter frame brightness continuity maintenance.The intra frame brightness adjustment method is the same as the above image enhancement algorithm.The returned illumination factor and histogram clipping point calculated from each frame are saved,The video scene is divided into three categories: background static scene,background gradient scene and background mutation scene by optical flow method,So that the visual quality of the final enhanced video sequence is improved while the brightness is continuous and stable without brightness flicker.This paper evaluates the enhanced image and video from both subjective and objective aspects,and selects four contrast algorithms to compare.The experimental results show that the proposed algorithm has better visual perception effect subjectively,and is better than other algorithms in the objective evaluation index score.