Pattern Based Image Retrieval System by Using Clustering Techniques

Nowadays, the image database is growing in enormous size with heterogeneous categories. Similarly, there is rising demands from users in various ways. The most demanding domains in society are health care, Agriculture, commerce, and security. Healthcare domain is concerned with diagnosing the disease. Security domain is involved in investigation of the Criminals. Commerce domain needs analysis to recognize the right product. Agriculture domain requires processing of disease affected fruit images. The PBIR system that combines evidence from multiple digital image domains can reduce those problems of existing image retrieval systems. The PBIR system is used to find uncertain parts of the image during augmentation steps. The result is identified with various parameters (e.g., accuracy, precision, distance, sum of distance), showing that the performance of the k-medoids pam algorithm better than the existing iterative clustering algorithms