Image Retrieval Based on Combined Texture with Shape Feature Using Wavelet
A new method combining texture with shape features using wavelet is proposed for image retrieval, which has the property of rotation, translation and scale invariance. The rotationnormalized image is achieved by definite the principal axis according to the weighted orientation of angular moments. The adoptive wavelet, which has the property of translation and scale invariant, is used to deal with the orientation-normalized image. The wavelet energy of every sub frequency band is extracted to be the texture feature of the image. Using the sub-graph rebuild the graph and then extracting the shape feature by moment invariants. Using the Gaussian model normalizes the texture and shape features. Finally, combining the texture features with the shape features retrieve. The experimental results show that the proposed algorithm has better robustness and precision.
image retrieval wavelet transform texture feature shape feature
Liu Li-ning Hou Yu-qing
School of Information Science and Technology, Northwest University Xian, China
国际会议
重庆
英文
293-296
2011-01-21（万方平台首次上网日期，不代表论文的发表时间）