The Research On Night Image Enhancement Algorithm And Its HLS Simulation

With the development of society,the status of monitoring equipment in our lives is becoming more and more important: police need monitoring equipment to catch criminals,drivers need driving recorders to maintain their legal rights,and stores need monitoring equipment to prevent theft.These monitoring devices ensure the quality of the picture under real-time recording.However,in the nighttime environment,serious problems such as low overall brightness of the image,loss of detailed information,severe color cast,and overexposure of local areas may occur on the screen of these monitoring devices.This thesis studies the nighttime image enhancement algorithm,and also researches on the high-level synthesis of the algorithm hardware implementation.Firstly,this thesis studies classic nighttime image enhancement algorithms: histogram equalization algorithm,dark channel prior theory algorithm,and Retienx algorithm.The basic principles of various algorithms are introduced,the simulation results of each algorithm are given,and the effectiveness and limitations of each algorithm are analyzed.Secondly,this thesis studies an improved night image enhancement algorithm based on Retinex.Introduce two important contents of the improved algorithm: bilateral filter and multi-scale analysis.Compared with the traditional Gaussian filter,the bilateral filter has a better effect of preserving image details.Multi-scale analysis also plays a role in enhancing nighttime images.The improved night image enhancement algorithm based on Retinex studied in this thesis is compared with other excellent night image enhancement algorithms.First evaluate the effects of various image enhancement algorithms from a subjective aspect,and then use objective evaluation indicators: average image brightness,image clarity,and image brightness sequence errors to evaluate various night image enhancement algorithms.At the same time,the possibility of implementing various night image enhancement algorithms on FPGA is analyzed.Finally,the high-level integrated HLS implemented by FPGA hardware is studied.HLS can use high-level language to write FPGA,and the written program has the parallel function of FPGA hardware description language,allowing FPGA engineers to design complex FPGA programs in a short time.Using HLS,an improved night image enhancement algorithm based on Retinex is implemented,and C and RTL level simulations are performed on it,and the error between the HLS simulation results and the theoretical results is small.Finally,the image result of FPGA platform implementing the algorithm of this thesis is given.