Embedded Real-time Image Processing Based On ZYNQ Under Different Lighting

Digital image processing system mainly processes images by computer.Image feature extraction is the hotspot of people's attention as the premise and foundation of complex image processing such as image segmentation,image matching,target tracking and computer vision.At present,the embedded image feature extraction is mostly processed by the pure software form of PC + image acquisition card.It takes a lot of time to process the image feature extraction algorithm,and the cost performance is not high.After that,the combination of single-core ARM,FPGA,and DSP caused the system to be complicated and did not receive good results.Based on the above series of problems,this paper proposes a new embedded image feature extraction system using Zynq chip development board.Zynq development board combines FPGA+ARM processor,adopts the combination of software and hardware to process the image feature extraction embedded system,and realizes the dual-core processing image function by using the combination of the two chip functions to complete a series of real-time images.The development of processing system design,Zynq uses the ability of FPGA parallel processing to realize the processing of Sobel algorithm,transplants the embedded Linux operating system based on Debian9 in ARM,mainly completes the data interaction with FPGA,and completes the control function of the whole system.The specific operations are as follows:(1)This paper performs edge detection processing on image feature extraction.According to the real-time requirements of the system and the complexity of the algorithm,the Sobel algorithm is chosen as the edge detection algorithm of the system.According to Zynq chip technology,the algorithm is packaged into HLS advanced synthesis tools.An IP core that can be applied in an embedded image processing system;(2)This paper adopts external OV5640 camera to realize image acquisition in FPGA,and uses VIVADO tool to build hardware platform based on Sobel IP core.The ultimate goal of this hardware platform is to realize image acquisition,image processing and HDMI high-definition display in one embedded system;(3)Based on the embedded hardware platform,the Linux operating system based on Debian9 is transplanted,and the real-time image processing of the image is finally realized.At the end of the article,the system was tested and analyzed on the MZ7XA development board.The built application was launched from the SD card and realized real-time image acquisition and processing.Then the static and software analysis and comparison of the software and hardware system processing results were compared.For the same 1280*720 image processing,the processing time of this system is about 3.4ms,and the software processing time is 33.2ms,the system is about 10 times of pure software processing in terms of image processing time,so it is constructed in practical engineering application.The embedded image processing system has great reference value.Finally,it is analyzed that different illumination has certain influence on the result of image acquisition and processing.Only reasonable control of illumination can make image processing have better effect.