投票表决算法在高帧频图像处理中的应用

Application of Voting Algorithm in High Frame Frequency Image Processing

针对目前高帧频图像处理方法中软件速度慢、实时性差、专用硬件开发周期长、灵活性差等缺陷,开发完成了基于FPGA的高帧频图像硬件实时处理系统。该系统采用投票表决算法,压缩了存储和处理的数据量,充分发挥FPGA器件的并行特性,使图像采集与图像处理并行完成,提高了图像处理速度。系统已成功应用于高速轨道检测车的钢轨断面图像实时动态处理和分析。

High frame frequency image processing at the present time has limitations, for software method is slow and lacks realtimeness, whereas special hardware has long cycle and bad flexibility. This paper presents a high frame frequency image hardware real-time processing system based on FPGA. The system adopts voting algorithm to implement synchronously image collection and operation, and reduce the amount of data that is needed to hold or handle. It exerts sufficiently the parallel characteristic of FPGA and observably improves speed of image processing, which has been applied successfully in real-time dynamic processing and analysis rail section of track inspection car.