摘要
采用Ｂ样条小波来提取层析图像的边缘轮廓，用高斯函数为平滑函数，用Ｍａｌｌａｔ快速算法进行小波分解，分解出的局部极值就是层析图像多尺度边缘。这两种算法在不同的尺度下有不同的提取边缘细节的能力和抑制噪声的能力。做到了抑制噪声和提取边缘细节之间有机的统一。
B-spline based wavelet transform is presented to detect the edge of the image, while the Gaussian function is used as a smoothing function and Mallat rapid algorithm is used to decompose the local maximum value, which is multiple scale of the edge. The results demonstrate that the wavelet transform used in this paper possesses different abilities to detect the edge and reduce the noise in different scales. The results also show that the wavelet transform is much better than the conventional differential operator based edge detection algorithms, such as Roberts, Sobel and Laplacian algorithms.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
1999年第11期1242-1245,共4页
China Mechanical Engineering
基金
国家经济贸易委员会重点产业化资助项目
关键词
逆工程
边缘检测
层析图像
小波变换
layer-by-layer 3D profilometry reverse engineering multiscale binary-wavelet edge detection
