Histogram Specification with Higher Order Polynomial Functions over R, G and B Planes for CBIR Using Bins

Main Article Content

Dr. H. B. Kekre
Kavita Sonawane


This work Proposes a histogram specification to modify the original histogram such that the intensities form lower level will get shifted to higher side which gives improvement in the results obtained for retrieval of images based on contents. Three polynomial functions proposed in this paper are designed and implemented for modifying the histogram of R, G and B planes of each image. These modified histograms are then partitioned into two parts using the center of gravity. Each partition has got id as ‘0’ and ‘1’. The three planes partitioned into two parts generating the eight combinations from 000 to111, which are used as eight bin addresses. These eight bins are holding the count of pixels having particular range of intensities based on the R, G, and B values falling in specific partition of respective plane’s modified histogram. Bins further are directed to have ‘Total of intensities’ and Average of intensities’ information of the image to be represented as feature vector. Total 21 feature vector databases are prepared by applying the feature extraction process to all 2000 BMP images in the database. Each feature vector in all databases is of dimension 8. This system is tested by comparing 200 query image feature vectors with all feature vector databases by means of the three similarity measures namely Euclidean distance (ED), Absolute distance(AD) and Cosine correlation distance (CD). Performance of the system is evaluated using three parameters PRCP (Precision Recall Cross over Point) Longest String and LSRR (Length of string to retrieve all Relevant images).


Keywords: Histogram Specification, Polynomial function, Bins, Count of Pixels, Total of Intensities, Average of Intensities, ED, AD, CD, PRCP, Longest String, LSRR


Download data is not yet available.

Article Details