Wavelet Edge Detection Based On Intra And Inter-Scale Dependency

Edge is a basic characteristic of image, which is important cause to slice and understand image。 Therefore, edge detection of image is a very difficult task on image analysis。 Because of edge detection complexity that edge and noise is high frequency signal, denoising and locating edge aren't very well。 So it is difficult to distinguish between noise and edge。It is known that wavelet transformation has good performance in local time domain and frequency domain。 Sub-images are acquired by multilevel wavelet transformations, then we can find that wavelet coefficients own similarity of statistical distribution in high frequency sub-images, that is to say, wavetlet coefficients distribute much wider through increasing scale of decomposition。 Sub-images of lower resolution whose wavelet coefficients own wider range, while sub-images of higher resolution whose wavelet coefficients own narrower range。 Therefore, wavelet coefficients mostly muster low frequency band。 But image edge and contour are high frequency imformation, which are detail information。 Its coefficients have more dependency through wavelet transformation, especially edge imformation in neighbourhood which own more cluster character in the same scale, also in the different scale, relevant wavelet coefficients have more dependency character。Therefore, the edge detection based on wavelet transform of inter-scale and intra-scale dependency will be presented。...