基于FPGA的颜色与运动特征识别系统设计

Design of color and motion feature recognition system based on FPGA

针对运动目标检测领域对实时性、便捷性、准确性的高要求,该文设计了一套基于FPGA的颜色与运动特征识别系统。系统辅以OV5640摄像头模块获取图像数据,基于Verilog语言进行并行流水线设计,使用SDRAM片外存储器实现数据流的实时缓存。基于FPGA实现了帧间差分法与HSV颜色空间特征的检测并进行结合,保证了系统的识别准确度。在FPGA中通过图像数据格式转换、运动目标检测、颜色特征检测、形态学滤波、目标指示跟踪等处理操作后实现了目标的精确识别。实验结果表明,该系统在不同的场景下,都可进行目标识别且识别准确率均达到95%以上。

Aiming at the high requirements of real-time,convenience and accuracy in the field of moving target detection,this paper designs a set of FPGA-based color and motion feature recognition system design. The system is supplemented by the OV5640 camera module to obtain image data,is based on Verilog language for parallel pipeline design,and uses SDRAM off-chip memory to realize real-time buffering of data streams. Based on FPGA,the inter-frame difference method and HSV color space feature detection are realized and combined to ensure the recognition accuracy of the system. The precise identification of the target is realized after processing operations such as image data format conversion,moving target detection,color feature detection,morphological filtering,target tracking and other processing operations in FPGA. The experimental results show that the system can perform target recognition in different scenarios and the recognition accuracy rate is above 95%.