Multiscale Product Riesz Wavelet for Color Edge Detection in Hilbert Domain

This paper proposes an edge detection technique based on the Multiscale Product of Riesz Wavelet Transform for color image. The product of the wavelet transform combines the advantages of the used scales: in large scale it is strongly anti-noise, and in small scale it can accurately locate edge. The Riesz wavelet transform is a complex wavelet transform (WT) which comes to overcome the discrete wavelet transform (DWT) shortcomings. The WT presents important additional properties. It is nearly shift invariant and directionally selective in two and higher dimensions. Computer simulations are performed on noise free as well as on noisy images. Comparisons with other edge detectors give improvement for the proposed edge detection technique.