摘要
针对目标检测系统对图像处理速度和信噪比要求高的问题,设计了一种基于现场可编程门阵列(Field programmable gate array,FPGA)的高速低噪声目标检测系统。首先,设计程序驱动CMOS摄像头传感器,采用同步动态随机存取内存(Synchronous dynamic random-access memory,SDRAM)进行实时存储,并利用乒乓存储实现存取速度的提高,然后对RGB像素数据进行灰阶化处理,提出一种由3*3快速中值滤波和4*4快速分权均值滤波相结合的双滤波算法,确保在减小数据边缘特征损耗的前提下去除系统随机噪声,通过Sobel边缘检测进行边缘特征提取,设计包围盒算法进行目标标定,最后通过视频图形阵列(Video graphics array,VGA)模块进行交互显示。测试结果表明,该系统具有图像处理速度快和检测精度高的优点,采用所提出的双滤波算法能比中值滤波、高斯滤波和均值滤波算法的平均信噪比高出约2.6倍。
Fast image processing speed and high signal-to-noise ratio are important to target detection system.Aiming at the problem,this paper designs a high speed and low noise target detection system based on field programmable gate array(FPGA).Firstly,a drive program is designed to drive CMOS camera sensor,synchronous dynamic random access memory(SDRAM)is used for real-time storage,and ping-pong Memory is used to improve the access speed.In order to remove the systematic random noise under the premise of reducing data edge feature loss,the RGB pixel data are grayed,and a double filter algorithm that consists of 3*3 fast median filter and 4*4 rapid separation of powers mean filter is proposed.The edge features are extracted by Sobel edge detection algorithm,and bounding algorithm for target calibration is designed.Finally,the video graphics array(VGA)module is used for interactive display.The test results show that the system has the advantages of fast image processing speed and high detection accuracy,and the average signal-to-noise ratio of the proposed dual filtering algorithm is about 2.6 times higher than that of median filtering,Gaussian filtering,and mean filtering algorithms.
作者
陈浩林
王擎宇
卢振坤
董艳珂
Chen Haolin;Wang Qingyu;Lu Zhenkun;Dong Yanke(School of Electronic Information,Guangxi University for Nationalities,Nanning 530000,China)
出处
《南京理工大学学报》
CAS
CSCD
北大核心
2022年第5期623-631,共9页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(61561008)
广西自然科学基金面上项目(2018GXNSFAA294019)
广西高校中青年教师科研基础能力提升项目资助(2020KY04027)
广西民族大学人才引进项目资助(2018KJQD18)。