Study And FPGA Implementation Of Color Image Enhancement Algorithm

With the continuous development of communication technology and microelectronic technology,the application of images and videos are more and more widely.However,due to the bad weather,poor light,and a bad camera,the captured images are usually dark and the details are lost,which seriously affects the visual effect and effective information extraction.By using the contrast enhancement method,the useful information of images can be highlighted and the image qualities can be improved.The processed images are suitable for subsequent application scenarios.Therefore,the study of contrast enhancement technologies has great scenes.In this thesis,we optimize and improve some classical enhancement methods.On this basis,we propose the total variation Retinex enhancement method based on texture suppression and the neighborhood histogram equalization.The research contents and results are as follows:(1)In this thesis,a total variation Retinex enhancement method based on texture suppression is proposed.Our method can avoid halo artifacts by constraining the illumination using L-1 norm.Besides,the details can be preserved almost.The results of the experiments showed that the subjective quality evaluation and the objective quality evaluation are both superior to the state-of-the-art enhancement methods.(2)A neighborhood histogram equalization method is proposed in this thesis.Neighborhood information is introduced into the histogram statics,which greatly avoids spikes in the histogram.On the other hand,the histogram is modified by adding the scale factor,which controls the enhancement level and improves the naturalness.The experiment results show that the evaluation results are superior to other similar methods.(3)An FPGA-based image enhancement system is built which utilizes Xilinx's Artix 7 series board.First,the neighborhood histogram equalization algorithm is synthesized into image processing IP core.Then a SoPC video processing system is put up by using software and hardware co-design method.Finally,1280*720 Video is enhanced in real time.