Research And Implementation Of Image Post-processing Enhancement

Recently, with the increasing popularity of portable electronic products such as the tablets and smart phones, a large amount of image data can be obtained by image acquisition at all times. However, the influence of the hardware limitation of electronic devices, the illumination of shooting scene and the technical level of the photographer during the data collection, storage or transmission process make the information involved in the image hard to be recognized, leading to the reduction of the image quality, and affects the further identification and determination.Consequently, these problems lead to the deterioration of the image quality. Therefore, this thesis presents a post-processing enhancement algorithm which can comprehensively improve color image quality and contains the brightness enhancement algorithm, texture enhancement algorithm and saturation enhancement algorithm:1. For the images with undesirable brightness, the thesis proposes a brightness enhancement algorithm based on multi-scale Retinex, which firstly converts the image to HSI color space, then enhances the brightness component of the image according to the reassembled multi-scale Retinex model, and finally uses a new linear quantitative method to process the Retinex data.2. In terms of clarity, an adaptive texture enhancement algorithm is presented. First of all, the algorithm uses gradient operators to detect the texture of the image, then selects its enhancement parameter adaptively on the basis of the distribution around each pixel as well as its gradient magnitude, and finally uses the fractional differential template to enhance the texture of the image.3. For color images, this thesis puts forward a saturation enhancement algorithm which combines the detection and the correction of the color cast. The algorithm can effectively enhance the saturation component of the image by means of the nonlinear transform theory of gray level and the result of color cast detection.According to the three above-mentioned algorithms and combining the characteristics of color space and the priority order of each image component, a relatively complete imagepost-processing enhancement algorithm is proposed. Based on the “Top-to-down” methodology, the algorithm is composed of five parts: the image classification module(color cast detection module), the color cast correction module, the saturation enhancement module, the brightness enhancement module and the adaptive texture enhancement module. Compared with several classic image enhancement algorithms, experimental results show that the proposed algorithm can not only improve the image quality based on their own information, but also maintain the color features and details of the images under the premise of ensuring enhancement. Furthermore, it has such excellent universality that the algorithm can be applied to enhancement processing for various types of images, such as normal,dark or colorful images.