Performing edge detection by Difference of Gaussians using q-Gaussian kernels

摘要: In image processing, edge detection is a valuable tool to perform the extraction of features from an image. This reduces amount information be processed, since redundant (considered less relevant) can disconsidered. The technique consists determining points digital whose intensity changes sharply. are, for example, due discontinuities orientation on surface. A well known method Difference Gaussians (DoG). subtracting two Gaussians, where kernel has standard deviation smaller than previous one. convolution between subtraction kernels and input results in this paper introduces extracting edges using DoG with based q-Gaussian probability distribution, derived q-statistic proposed by Constantino Tsallis. To demonstrate method's potential, we compare introduced tradicional kernels. showed that extract more accurate details.