线状目标实时检测算法的研究

提出了一种基于FPGA和DSP的算法并设计了相应的硬件系统,能对大小为256×256像素、帧频为25帧/s的视频图像进行实时处理。即先对输入图像采用改进的Sobel算子提取边缘,再用改进的Hough变换算法进行直线检测。解决了传统的Sobel算子提取的边缘较粗及Hough变换算法计算量大、难以实时实现的难题。

Research on real-time detecting algorithm of linear objects

A new algorithm based on FPGA and DSP was stated,and the corresponding hardware was designed. With this system,the video images,with the frame frequency reaching 25 and the size reaching 256 × 256 pixels,can be processed in real-time application. The procedure can be divided into two steps. Firstly,the edge of the input image was extracted with the improved Sobel operator. Then detection of lines was done to the image with the improved Hough transform algorithm. The conventional Sobel operator works not so well on edge extraction because the edges extracted with it are very thick,and the conventional Hough transform algorithm is disadvantageous in computation price so that it can't be implemented in real-time systems. With the improved algorithms and hardware system,the problems were resolved.