Color extraction of textured images using discrete wavelet transform

Color extraction of textured images is widely used in a variety of image-related functions. In this study we present an efficient algorithm to completely separate the original color of homochromatic images with different texture structures. 63 different samples in 9 colors and 7 textures have been examined by three texture analysis approaches including statistical methods and discrete wavelet transform in modified HSV color space. In wavelet transforms frequency domain by resetting the detail coefficients of images in HSV color space to zero (which contain high frequency texture features) and reconstructing to spatial domain, an average of 0.93% in standard deviation of reconstructed color intensities is obtained, while statistical methods have illustrated 2.79% of standard deviation in average.