Content based Image Retrieval based on Cumulative Distribution Function - A Performance Evaluation

Content Based Image Retrieval (CBIR) is very much sought after field in the area image retrieval. CBIR finds applications in areas such as web searching, crime detection, military, intellectual property and medical diagnosis. With the advancement of medical imaging modalities, focus on the diagnosis has shifted from physician centric to hospital based diagnosis where, image analysis & interpretation is the key factor in diagnosis. With the image acquisition becoming digital there is greater need to interpret the medical images in quick and accurate manner. In this paper an image matching technique based on Cumulative distribution Function (CDF) for retrieving the medical images from the database is discussed. This method can provide considerable reduction in the image retrieval time while providing flexibility to the physician. The physician can select suitable number of CDF line segments for comparison and the percentage of CDF threshold as desired by him there by providing control in terms of Precision (P) and retrieval time (Tr).