摘要
针对传统Sobel边缘检测算法需要人为设定阈值,且对图像复杂纹理的识别能力较弱等问题,文中提出一种图像自适应阈值及3×3算子的8方向Sobel边缘检测算法,并在现场可编程门阵列(FPGA)上通过流水线设计和乒乓操作加速实现。基于Xilinx公司的Spartan6系列FPGA硬件平台,将OV5640摄像头采集到的图像数据流缓存于SDRAM;再利用移位寄存器将图像数据传输至图像算法模块,经过图像中值滤波及8方向Sobel边缘检测算法进行处理;最后,通过驱动VGA实时显示视频图像的边缘信息。实验结果表明:利用文中算法并基于FPGA输出的图像,在有效降噪的同时能够较好地滤除虚假边缘并保留真实的边缘信息;相较于传统算法,利用所提算法处理后的图像边缘清晰度有明显提高,且系统具有良好的准确性和鲁棒性。
As the traditional Sobel edge detection algorithm requires thresholds to be set artificially and has weak capability to recognize complex textures in image,an 8⁃direction Sobel edge detection algorithm with image adaptive threshold and 3×3 operator is proposed,which can be accelerated on FPGA(field programmable gate array)by means of pipeline design and ping⁃pong operation.The image data stream collected by the OV5640 camera is buffered in SDRAM on the basis of hardware platform of FPGA in Xilinx′s Spartan6 series,and the shift register is used to transmit the image data to the image algorithm module,which is processed by means of image median filtering and 8⁃direction Sobel edge detection algorithm.The edge information of video image is displayed in real time by driving VGA.The experimental results show that the algorithm can filter out false edges and retain the real edge information of the image based on FPGA output while effectively reducing noise.The comparative experiments show that,in comparison with the traditional algorithm,the edge sharpness of the image processed by the proposed algorithm is significantly improved,and the system has better accuracy and robustness.
作者
柯青林
张文涛
余英铨
方博闻
KE Qinglin;ZHANG Wentao;YU Yingquan;FANG Bowen(College of Electronic Engineering and Automation,Guilin University of Electronic Science and technology,Guilin 541004,China)
出处
《现代电子技术》
2022年第24期92-97,共6页
Modern Electronics Technique
基金
国家科技重大专项课题(2017ZX02101007⁃003)
国家自然科学基金项目(61965005)
广西自然科学基金项目(2019GXNSFDA185010)。
关键词
边缘检测
自适应阈值
8方向Sobel
数据传输
中值滤波
图像降噪
edge detection
adaptive threshold
8⁃direction Sobel
data transmission
median filtering
image noise reduction
