【摘要】 提出用于图像边缘检测的小波包分解算法.首先对图像的高频子空间进行多层次小波包分解,接着求出各层同一子空间不同方向上系数的差值,然后分层对差值进行阈值化处理,最后依据每层阈值化后的系数,得到不同分辨率下图像的边缘.该算法不需预先对图像进行去噪处理,却能很好地抑制噪声.实验结果表明,该算法既能根据需要分别求出不同尺度下图像的边缘,又能保持图像边缘的完整性和清晰性.更多还原

【Abstract】 The algorithm on edge detection of an image base on wavelet packet decomposing is put forward. High-frequency subspace of an image is first decomposed into multi-level wavelet packets. Second, the coefficient differences are calculated on different directions of the same subspace at all levels. Third, the threshold differences are dealt with at each level. Finally, the edges of images of different clarities are obtained based on the threshold coefficient at each level. Based on this algorithm, noise restrain does not need to deal with in advance. Good results are obtained in noise restrain. Results show that the algorithm can not only calculate the edge of an image at different standard, but also keep it complete and clear.更多还原