Image pattern recognition by edge detection using discrete wavelet transforms
Ravikant Divakar, Bijendra Singh, Ashish Bajpai, Anil Kumar · Journal of Decision Analytics and Intelligent Computing · 2022
An edge is the high-frequency part of an image and represents the location where abrupt changes take place in the intensity of luminescence. Edge detection is a basic step in feature extraction and pattern recognition of any image. Wavelet transforms extract low- and high-frequency information from any signal separately. In a two-dimensional wavelet transformation, an image is decomposed into four sub-images: one approximation image and three different images (horizontal, vertical, and diagonal images) at each decomposition level. The difference images show how the neighboring pixels differ in the horizontal, vertical, and diagonal directions. The approximation coefficients are forced to zero, and the difference coefficients are inverse wavelet transformations, as the reconstructed image shows the edges of the image and describes its pattern. Using the Haar wavelet at decomposition levels 1, 2, and 3, image pattern recognition by edge detection is performed and discussed.