Image Classification using Wavelet Watermarking Representation

In this paper, a novel approach using wavelet watermarking representation for content-based image classification is proposed. The original image is decomposed into wavelet coefficients using discrete wavelet packet decomposition. The energy signatures of most dominant channels are extracted as the image feature, and encoded as a watermark. For minimal loss in image fidelity, the watermark is then embedded in those wavelet coefficients with highest magnitudes except for those in the lowest frequency channel. The image can be retrieved by a simple watermarking detection method. Experimental results show that the proposed method is effective and achieves the best performance of 94.8% accuracy for image classification.