基于改进Hough变换的圆形物体检测

圆形(包括弧线)检测是数字图像处理过程中的经典问题之一,有着广泛的应用背景。基本的Hough变换方法是将图像中的每一边缘点映射到参数空间的一个区域,选取累积最多的参数。在现实生活中,由于噪音、数字化错误和图形变异等因素真实的图形经常被曲解,因此,图像在应用Hough变换后,很难找到单一的峰值,这也就造成了检测的难度。在Hough变换的原理基础上利用圆的几何特征提出了改进算法。理论和实验证明能获得较好的检测结果。

Circle Detection based on Improved Hough Transform

The detection of circle including camber is one of classical problems in digital image processing, and has extensive application background. The classical Hough transform method is to map each edge point of image to a region of parametric space, and select the most accumulated edges, In real-life images, the shapes are often distorted from their true parametric forms due to the presence of noise, digitization error and shape variations. Therefore, after applying Hough transform, it may be difficult to find out a single peak, thus it is difficult to detect image. On the basis of Hough transform, this paper presents the improved algorithm by using the geometric characteristics of circle. Theories and experiments show that it has better detection performance and can get accurate results.