Research And Implementation Of Real-time Image Preprocessing Technology Based On FPGA

With the rapid development of economy,people's awareness of safety is constantly improving,meanwhile,due to the need of safe city construction,the demand for video surveillance equipment continues to grow.Video surveillance is playing an increasingly important role in our daily life,combined with image processing algorithm,target recognition and tracking can be realized,the image preprocessing technology directly affects the accuracy of target recognition and the quality of video surveillance,so it is necessary to conduct in-depth research.The video surveillance system has the application requirements of low power consumption,portability,high speed and low cost,FPGA has the characteristics of high speed,parallel operation and strong flexibility,so it can meet the design requirements of video surveillance system.So this thesis uses FPGA as the main controller to study the real-time image preprocessing technology and carry out the engineering implementation.The main research contents of this thesis are as follows:(1)In the video monitoring equipment,the image is interfered by the external environment,electronic devices and data transmission,which is easy to produce noise,thus causing bad influence on the imaging results.Therefore,the image should be denoised first.Aiming at the problem that median filtering effect is easily affected by the size of filtering window,an improved adaptive median filtering algorithm is proposed in this thesis.By fixing the maximum window size,the image gray extreme value in the window is judged by noise,and the pseudo noise points are removed.When the final filtering effect does not meet the output conditions,the maximum number of pixels in the window is used to replace the median filter for output.(2)The purpose of edge detection is to detect the discontinuous changes in the gray area of the image and determine its precise position,so as to provide information for imaging analysis.In view of the fact that the Sobel operator needs to set the threshold manually,an improved adaptive Sobel operator is proposed in this thesis.By selecting the mean value of the most relevant pixels in the neighborhood of center pixel 4 as the threshold value,the real-time edge detection is improved.(3)This design takes FPGA master chip as the core,designs the hardware circuit of the whole system,and implements the improved adaptive median filter algorithm and adaptive Sobel algorithm.Through image comparison,the results show that the improved adaptive median filtering algorithm and adaptive Sobel algorithm can make the image clearer and the edge more detailed,and realize the design of real-time image preprocessing system.