Design of an IOT Based hybrid and intelligent content Based image retrieval system

Abstract

Images play a vital role in many applications related to academia and industry. There are numerous applications of images in various prominent domains like face recognition, finger print matching, medical diagnostics, crime prevention etc. With the tremendous advancements in image capturing devices like portable cameras, sophisticated mobiles, etc., there has been a cumulative increase in the number of images. But, manual searching, managing, indexing and retrieving some particular images from large storehouse of images is a tedious task. Sometimes, some sort of textual annotation is also being added to images for the ease of users. But, manual annotation by some operator or user itself is a difficult and time consuming task. Also, skilled users are required for the annotation task. Image retrieval systems in which images are retrieved on the basis of textual annotation are designated by Text-based Image Retrieval systems (TBIR). But, TBIR systems are prone to many types of errors like misspellings, homonyms, synonyms, etc. Therefore, the key to the above issues and for the successful retrieval of images, Content-based Image Retrieval (CBIR) came into existence. newlineCBIR system is based on the extraction of visual attributes of an image like color, texture, shape, spatial information, etc. These features are categorized as the low level features of an image and are very close to human visual perception. In this thesis, a hybrid CBIR system is developed which is based on an amalgamation of color, texture and shape based low level features. Color Moment, Gray Level Co-occurrence Matrix (GLCM) and Region Props are used here for the extraction of color, texture and shape features respectively. newline