Content Based Image Retrieval: A Study of Approaches and Techniques

The Content Based Image Retrieval (CBIR) is basically used to extract images from the collection of images using several image matching and extraction methods. As it is now with the advancement and ongoing consistent technology and negligible cost of the electronic, digital and virtual sensors, a huge collection of image databases is processed by different types of applications. This type of database of images extends the requirement that we should provide effective search methods for the retrieval of images that can fulfill the demands of end-user. The texture, color, or shape of a picture is the most basic feature used to symbolize and index the images. These distinct elements of photographs are retrieved and used as a measure to find image similarities. A search procedure is initiated for relevant images that fulfill the user's information need using a sketch drawing or a query image given by the user that has similar features. The relevance feedback can be used as an optional step to retrieve and enhance the CBIR process. In this paper, after extracting semantic characteristics from photos, an analysis of the content of the images is performed with respect to features connected to the low level. In this paper, an overview of traditional CBIR methods (e.g.: Color histogram-based, texture based and shape based) are discussed. This paper's primary goal is to study the comprehensive details on the various computational techniques used for content-based image retrieval. To further enhance the efficiency of the image retrieval process, their limitations are also addressed.