摘要
角点是图形图像的重要特征,对图形图像的理解和分析具有重要作用.角点检测的关键是对局部支撑域和曲率的估算.为此,提出一种基于直线拟合的自适应角点检测算法,设计了滑动窗口自适应调整策略,并用于边界点局部支撑域的选择.克服了以往局部支撑域事先固定不变的缺陷,且在滑动窗口的基础上提出基于特征向量夹角的曲率计算方法,克服了图像边缘噪声的干扰.针对改进的Harris角点检测算法和n链码角点检测算法进行比较,实验证明所提出的算法能对角点进行精确定位,抗噪能力强.
Corner is one of the most important features of image. The key of corner detection is to estimate the region of local support and curvature, so an adaptive corner detection algorithm was provided based on linear fitting. And the gliding window adaptively-designed was used to calculate the region of local support and over- come the disadvantage of the invariable region of local support. In addition,a curvature calculation algorithm was also provided based on eigenvector within the gliding window to overcome the distraction of noise. Compared with Harris corner detection algorithm and n-chain code corner detection algorithm, contrast experiment shows that the new approach is robust and accurate.
出处
《上海工程技术大学学报》
CAS
2009年第1期46-50,共5页
Journal of Shanghai University of Engineering Science
关键词
角点检测
最小二乘法
协方差矩阵
corner detection
least squares method
covariance matrix
