Information retrieval for unstructured multimedia in web mining

Abstract

Visual Content Information Retrieval technique mines required multimedia data newline(i.e. videos, images) from a large collection of database based on visual contents. The newlineVisual Content Information retrieval is one of significant research areas in the field of newlineimage and video retrieval. Many research works has been intended to perform visual newlinecontent information retrieval with helps of different data mining techniques. The newlineperformance of existing visual content information retrieval techniques was not efficient. newlineIn addition, Multimedia information retrieval is important to retrieve multimedia data newlinerapidly. A lot of research works were designed for multimedia information retrieval with newlineassists of classification techniques. However, performance of conventional classification newlinetechniques was not sufficient for achieving improved precision and recall. There is a newlinerequirement for effective retrieval system to retrieve relevant video contents from a large newlinedataset according to user query. newlineExisting Histogram Clustering Technique was designed to mine video data from newlinevideo database. The Histogram Clustering Technique lessens the time taken for video newlineretrieval. However, retrieval performance of Histogram Clustering Technique was not at newlinerequired level which lacks precision and recall. Further, an Automatic Setting Similarity newlineThreshold (ASTS) was developed to perform content-based visual information retrieval. newlineThe ASTS enhances the precision and recall of video retrieval. The time complexity of newlinecontent-based visual information retrieval was higher. Moreover, Existing Large-Scale newlineVideo Retrieval was intended to carry put scalable and memory-efficient video retrieval. newlineHowever, precision and recall rate of video retrieval was poor. Besides, conventional newlineContent-based video retrieval was developed to retrieve human actions videos from video newlinedatabases. The performance of content-based video retrieval was not effective. In order to newlineovercome the existing issues, three proposed methods namely GSTC technique.