Based On The Traffic Lights Control Design Of Low-quality Image Enhancement Method

In the process of urbanization in China today,traffic congestion has become an important problem that cannot be ignored in people’s life.With the development of data acquisition technology and information technology,intelligent transportation system has become an inevitable choice for traffic and transportation management.At present,most of the studies based on outdoor machine vision traffic light control system do not consider the influence of complex environmental factors,such as fog,low illumination,etc.,under such interference,it is difficult to carry out effective system processing of acquired images.The current image processing is basically carried out in the background,the server needs to calculate a large number of images to be processed every day,causing the response delay,so it is necessary to move part of the calculation to the front end,the application of embedded technology in the field of intelligent traffic light control has become an important means.This topic focuses on the machine vision processing of a phase traffic light control system at a crossroads.In order to relieve the pressure of backstage server and manual operation,the operation of image enhancement and vehicle recognition are all carried out under the embedded system,and the operation results are used as the effective premise of intelligent control of traffic lights.In the part of low-quality image enhancement,this topic studies the foggy degradation model and the dark channel prior dehazing algorithm for foggy image enhancement.According to the theoretical similarity,the optimization idea of single image fast dehazing algorithm is proposed.The brightness self-adaptation problem is deeply studied and the brightening parameters in the dehazing algorithm are analyzed.Finally,the V component processing of HSV and RGB conversion model is used to improve the overall brightness of the image.For the enhancement of low illumination image,a global tonal mapping fusion algorithm based on two-bit gamma function is proposed.The tonal mapping theory is mainly studied,and the luminance correction is carried out through two-bit gamma function to the uneven illumination phenomenon at night.Through subjective and objective evaluation and comparison of the simulation results of the traditional image enhancement algorithm and the proposed algorithm,it is concluded that the proposed algorithm can enhance the image information features better.In this topic,STM32 is used as the embedded control system chip,and the effectiveness and real-time performance of the defogging algorithm and the low illumination processing algorithm are verified by MATLAB.In order to realize embedded image processing,this topic proposes two kinds of program transplantation methods,which are using Matlab Coder to transform C language for program transplantation and Open MV calling program transplantation in Keil environment respectively.Finally,the identified vehicle targets are counted with pixel statistics method.The design of traffic light control system based on STM32 is provided.