Image Based Search Engine with Deep Learning

Abstract: The objective of this paper is to present a brief overview of existing Image Based Search Engine(IBSE) technique. The IBSE method is used to retrieve relevant images from the database based on the query image submitted by the user. The retrieval of images from a database relies purely on the image features such as color, shape and object identification using texture(s) in the query image. The CBIR area is more diverse as it is used in different domains like medical image diagnosis, classification and face recognition etc. The objective of this paper is to analyzing and discussing the latest developments in this field. Comparison contains the methodological analysis along with the prospective advantages and disadvantages. The results discussion of the previous methodologies has been analyzed also. It provides the gap identifications and future challenges which can be redirected in the near future. This discussion is based on the retrieval techniques which used the color, texture, shape and dominant color in the image retrieval.

Keywords: IBSE, Feature Extraction, Feature Dimension, Dimensionality Reduction, Relevance Feedback, Similarity Measures.