Spectral Correlation & Combined Moments Framework for Application of CBIR at Biomedical Database

A new feature extraction technique for Content based retina biomedical image retrieval is proposed. This method is based on spectral correlation that provides description in the frequency domain. The image is partitioned into non-overlapping tiles. The features drawn from transferred image with proposed new features Spectral correlation using moments between the image tiles, serve as local descriptors. Shape information is captured in terms of edge images computed using combined moments. The combination of the texture features between image and the shape features provide a robust feature set for biomedical retina image retrieval. The experimental results show the efficacy of the method. For matching the biomedical retina images an integrated matching, based on similar highest is provided. The experimental results are compared and found to be encouraging.