Application of Multi-scale Gabor Wavelet Transform in Image Retrieval
Zhu Mingzhong · Electronic Science and Technology · 2011
Texture is a very important feature in content-based image retrieval,which is difficult to describe.To improve the efficiency of texture feature extraction in image retrieval,the structure of Gabor filter banks is studied,and an approach to texture images retrieval based on multi-scale Gabor wavelets transform is proposed.A group of Gabor filters with multi-scale and multiple directions is designed,and the parameters are optimized before image filtering and feature extraction.An image retrieval system based on Gabor texture feature is designed,and some image retrieval experiments are curried.The results show that this method based on multi-scale Gabor transformation has an excellent performance on texture extraction.