一种改进的Sobel边缘检测算法的设计及其FPGA实现

为了快速得到边缘更清晰、敏感度更高的实时边缘图像数据,提出了一种改进的Sobel边缘检测算法.对原始图像的边缘进行平滑滤波后,利用2个3×3模板对选定的二维图像中同样大小区域内的数据进行卷积,合并水平及垂直方向的导数得到该区域的梯度值,再通过添加的梯度方向模块对该区域进行4个方向的判断以帮助确定边缘,最后通过非极大值抑制及二值化处理确定图像边缘.将改进后的Sobel边缘检测算法在FPGA上进行了实现并证明了其有效性.

Design and Realization of Improved Sobel Edge Detection Algorithm Based on FPGA

An improved Sobel edge detection algorithm has been proposed to obtain the real-time edge of image data quickly with sharper edge and higher sensitivity.After applying the smoothing filter to the edge of the original image,two 3×3 templates are used to perform convolution on the data of the same size region in the selected two-dimensional image.The gradient values of this region can be obtained by combining the horizontal and vertical direction derivatives.Then,the gradient direction module is used to help judging the four directions of this region and defining the edge.The image edge is finally determined by non-maxima suppression and binarization.The improved Sobel edge detection algorithm is realized based on the FPGA and proved to be effective.