A Hybrid Image Mining Technique using LIMbased Data Mining Algorithm

Abstract

ABSTRACT The field of image retrieval and mining has become a vibrant research area due to speedy enhancement in the volume of digital image databases. Nowadays, a large portion of information is in visual form; it is essential and certainly pleasing to search for images by content. Image mining has a variety of applications in various sectors like medical diagnosis, biology, remote sensing, space research, etc. This research is to determine the exact images while mining an image (multimedia) database and proposes a novel approach for mining images using LIM based image matching technique with neural networks. This process is independent of too many parameter setting to generate a robust solution. It is designed and implemented on MATLAB and is tested with the images of various databases. Appropriate measures were devised to evaluate the performance of the system. The performances of the LIM based image matching technique results were noteworthy and comparable. While comparing with the number of false retrievals with the correct retrievals, the anticipated system performance level will be suited for several simple day to day multimedia database applications and image mining systems. The image mining system derived from the LIM based image matching technique provided promising results.