Performance Evaluation of Bins Approach in YCbCr Color Space with and without Scaling
H. B. Kekre1, Kavita Sonawane2

1H. B. Kekre, Computer Engineering, NMIMS University/MSTME/ Vile Parle, Mumbai, India.
2Kavita Sonawane, Computer Engineering Department NMIMS University /MSTME/VileParle, Mumbai,India.
Manuscript received on June 03, 2013. | Revised Manuscript received on June 27, 2013. | Manuscript published on July 05, 2013. | PP: 203-210 | Volume-3 Issue-3, July 2013. | Retrieval Number: C1711073313/2013©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: This paper explores novel idea of feature extraction based on bins approach. The bins formation process described in this approach is based on the partitioning of the different histograms of the image based on the color components of the image. Basically the feature extracted here deals with the color contents of the image. The use of color contents is explored using two different color spaces namely RGB and YCbCr color space in two forms with and without scaling. Feature extraction phase starts with the separation of the color planes of the image. In this work images in both (RGB and YCbCr) color spaces are separated into R, G, B and Y, Cb, Cr components respectively. Histogram for each plane is calculated and partitioned into two parts using Centre of Gravity (CG) technique. Three planes, two partitions generate total 23 = 8 bins. Data contained by 8 bins is the count of pixels falling in particular range of intensities. This is further processed by computing the first four moments. It generates four types of feature vectors based on four moments namely Mean, Standard Deviation, Skewness and Kurtosis. Feature vector comparison with query is facilitated by means of three similarity measures namely Euclidean distance, Absolute distance and cosine correlation distance. Experimentation is carried out using 2000 images in the database with two color spaces RGB and YCbCr taken into consideration. Result analysis is done by using three performance evaluation parameters Precision Recall Cross over Point, Longest String and Length of String to Retrieve all Relevant images.
Keywords: Bins, CG, Histogram, RGB, YCbCr, Mean, Standard deviation, Skewness, Kurtosis, ED, AD, CD, PRCP, LS, LSRR.