VLSI realization for decomposition of dynamic texture using efficient motion estimation algorithm

Dynamic textures are sequences of images of moving scenes that exhibit certain stationary properties in time, these includes sea-waves, smoke, foliage, whirlwind, crowd, etc. This proposal is made to present a novel characterization of dynamic textures. That poses the problems of modeling, learning, recognizing, classifying, synthesizing, and editing dynamic textures on a firm analytical footing. Dynamic textures are decomposed using diamond search motion estimation methodology; Diamond search motion estimation algorithm which could divides the frames into 16×16 pixels that can make the decomposition faster when compared to existing MCA method. The RGB frames are converted into gray scale image using MATLAB tool and resulted MIF (memory initialization file) is applied to diamond search logic which is designed using MATLAB. Our proposed system increases the processing speed of decomposite textures. The Morphological Component Analysis (MCA), and Diamond Search (DS) algorithm are used calculate the PSNR values.