Research On Real-Time Edge Detection Of Foggy Traffic Image Based On SOPC

Edge computing migrates cloud service to the edge of network by introducing Edge-Device between terminal device and cloud service.With regard to transportation,Edge-Device is urgently need to reduce network burden and storage capacity.Because of the outdoor traffic imaging equipment produce increase,brings about the data sharp increase.Besides,affected by impurities in the air,image are degraded seriously and the definition of images are very terrible.On this foundation,the article takes traffic image in the fog or haze weather as target,integrating filter denoising,image defogging,edge detection and morphological operation into one Intellectual Property(IP)core,to achieve the real-time edge detection system of foggy traffic image on System on a Programmable Chip(SOPC)development platform.The article regards Zynq-7000 development platform as Edge-Device.uses the high-level synthesis tool Vivado HLS,which introduced by Xilinx.It designs the IP core of traffic image defogging and real-time edge detection,to realize the hardware acceleration of it.Besides,using the advantage of SOPC technology and software co-processing and combining now available IP core to build the system of foggy traffic image real-time edge detection.This article mainly studies the following aspects:(1)Analyze the method of image filtering and image defogging and edge detecting and morphological operation,find the commonality of these types of image processing algorithms to integrate together.The color image were split into three components to processing separately,according to the features of HLS and the requirements to mix modified three channels together.(2)By studying the features of HLS design and foggy image and its histogram,according to the design and research,the image filtering and histogram equalization and Sobel edge detection and morphological operation were implemented into one IP core by Vivado high level synthesis tool,which achieved the purpose of defog and edge detection.(3)By studying the optimization measures and strategies of Vivado HLS tool and aim at real-time,comparative analysis of different schemes,and consider resource consumption and time efficiency into to choose the most suitable scheme,then export IP core.(4)By using Vivado tool,and combine ready-made IP cores and custom IP core to build real-time edge detection system of foggy traffic image,and then driving and debugging the system by Vivado SDK.Finally,the system verification is performed on the Zedboard development board,the performance of edge detection is evaluated by quality factor of FOM,and the system performance is analyzed from power consumption and timing and resources according to the synthesis report.The experimental results show that the SOPC implementation can satisfy real-time edge detection requirements.