Construction of NPR biorthogonal wavelet and its application in edge detection
Qiang Zhu, Lao Qicheng · 2010 3rd International Congress on Image and Signal Processing · 2010
Based on parameterized algebraical method, the non-prefect reconstruction (NPR) biorthogonal wavelet is presented. According to the particularity of image edge detection process, the perfect construction condition is relaxed. The low pass decomposition filter of non-prefect reconstruction biorthogonal wavelet is even symmetry about zero location, so that the edge position is more accuracy in multi-scale decomposition. The high pass decomposition filter is odd symmetry about 1/2 location and the sign is monotonic. The wavelet with above properties can ensure the validity of edge detection.