Texture Modeling and Segmentation towards content based image retrieval system


There are many techniques to extract features. Some use direct and other use transformation such as Fourier Transform, Discrete cosine transform and Wavelet Transform. Such Transforms convert the signal from time domain to frequency domain and vice versa. The Wavelet Transform conducts the change a frequency and reflects that in time. This research uses the Wavelet Transformation method to process images with (128*128) pixels, knowing that these images are in grayscale. An algorithm is proposed for texture classification using Wavelet Transform. The texture classification of an image means dividing it into sub images of fixed or variable length. Hence, this algorithm first takes the image and then it divides it into block of (8*8) dimension. There are two phases mainly in the proposed algorithm, classification and retrieval phases. In the classification phase, the procedure that followed requires the texture segmentation and wavelet transform computation of the test image. In the retrieval phase, the classified texture will be labeled following a given logical decision.