Image Decomposing by Discrete Wavelet Transform in the Image Retrieval Systems

In this paper, we propose a CBIR method that uses wavelet transformation. The property of wavelets to localize both time and frequency makes them very suitable for analysis of nonstationary signals [1]. They are an excellent tool for feature extraction, signal and image compression, edge detection and compression. The reason of using the wavelet transform is that the basis functions used in wavelet transforms are locally supported; they are nonzero only over part of the domain represented. Hence, adequately chosen wavelet basis groups the coefficients in two groups – one with a few coefficients with high SNR, and other with a lot of coefficients with low SNR. Using the wavelet coefficients of images we compute a pseudo-hash information that is later used for fast querying the database. This approach for searching an image database in which a query is expressed as a low-resolution image is known as query by content [2]-[5].