Content based image retrieval using modified histogram and wavelet with user feedback system

Abstract

The CBIR is an effective method to automate the process of image retrieval in an effective manner. The retrieval of images from the database is carried out with features and classification algorithms. The retrieval efficiency in retrieving an image depends entirely on the features. It is important to utilize the image database in an effective way and making it robust against retrieval of images. Since, the image diversity has been increasing in these databases, the processing of retrieval is increasing rapidly and hence the content based image retrieval gains huge attention by the researchers. The most problem the color content based image retrieval system face is the lack of spatial distribution vectors. The present study retrieves the image from large database using correct spatial information. This is achieved with the use of wavelet function, which accurately predicts the spatial information for the extraction of color feature points. The use of color distribution generates the vector of color distribution and the use of wavelet features helps in generating multiple image layers that helps in finding the spatial distribution of color vectors. In the first part, the present study uses Multi-layer Multi-level Color Distribution (MLMLCD) approach in image retrieval algorithm. This method uses wavelet transform that reduces the signals with less distortion and the images are transformed to different levels by varying the wavelet parameters. The color distribution vector is generated over each wavelet transform output and produces three different distribution vectors for each level. newline