Enhanced Retrieval Accuracy with ZMs using Dual Tree Complex Wavelets and Fourier Features

Though Zernike Moments (ZMs) have been most widely used in extracting the region based shape features of an image related to Logo, Trademark, Clip art search etc. Still retrieval of accurate images is an issue. It extracts global features and ignores local details. To overcome this drawback, wavelets can be used along with ZMs. In this paper, Invariant Dual Tree Complex Wavelets and Fourier Features (DTCWT-FDs) are applied with ZMs in order to give composite view of local and global descriptors. Firstly features using ZMs and DTCWT-FDs are extracted for whole image database and stored separately. Afterwards query image is given and our system retrieves images only on the basis of features of ZMs as it is a region based descriptor. With this the irrelevant images are filtered out and the selected database after filtering will be compared with query image on the basis of features of DTCWT-FDs. In DTCWT-FDs, image is transformed from Cartesian coordinates into polar coordinates. Fourier Transform (FT) is applied on them. Afterwards the Fourier coefficients are fed to Dual Tree Complex Wavelet Transform which due to its good directionality property captures better local details of images. Here local and global features are extracted with DTCWTFDs and ZMs respectively in two stages. The above said two stage hybrid technique improves the retrieval efficiency of the system. Region based as well as contour based Shape databases are used in the study. Due to the characteristics of this technique, these features are rotationally invariant.