WITHDRAWN: Image Digest using Color Vector Angle and Dominant Walsh-Hadamard Transform Coefficients

Abstract

Classical image digests often investigate the luminance plane (Y) in the YCbCr color standard to produce robust hash values. However, in most cases, the images are in RGB format. So digests employing luminance planes suffered from poor discrimination. The proposed work develops a robust image digest for RGB images, which exploits all R, G, and B planes of images to obtain intense discrimination of the digest. The proposed digest reprocesses images in the database to a fixed dimension. Later, it extracts dominant approximation components by using a Gaussian Low pass filter from each color component R, G and B. Thirdly, Color Vector Angle (CVA) is applied on merged three planes of the image. Fourthly Walsh-Hadamard transform (WHT) is used on 8 × 8 blocks. The dominant dc coefficients of WHT are taken for the hash generation. Finally, it takes the Hamming distances as a similarity measurement between the original and extracted hashes. The simulation results indicate that our proposed algorithm is superior to the classification efficiency among perceptual robustness and discriminating capability.