Please use this identifier to cite or link to this item:
http://hdl.handle.net/2080/2476
Title: | A Novel Image Enhancement Technique based on Statistical Analysis of DCT coefficients for JPEG Compressed Images |
Authors: | Bhatia, J Okade, M |
Keywords: | Image Enhancement Technique DCT JPEG Compression |
Issue Date: | Mar-2016 |
Citation: | 22nd National Conference on Communications(NCC 2016), Guwahati, Assam, 4-6 March 2016 |
Abstract: | This paper investigates a novel image enhancement technique based on statistical analysis of DCT coefficients. The proposed method treats the DC coefficient and AC coefficients of a DCT block separately to arrive at two different scaling factors. Since the DC coefficient of a block being an average value concerns with illumination of the image, the coefficient of variation (cv) is estimated which is used as the DC scaling factor. On the other hand, since the AC coefficients concern with the local contrast, these are analyzed diagonally forming bands followed by calculating entropy and contrast which are utilized to arrive at the AC scaling factor. The two scaling factors are used to compute the enhanced DCT block and the corresponding enhanced image. Our experimental results show that such statistical analysis is able to capture the behaviour of the DCT coefficients which is exploited to arrive at the scaling factors thereby making the proposed method independent of predefined enhancement functions like twicing function, S-function currently utilized by state-of-the-art methods. Extensive testing and comparative analysis of the proposed method clearly shows better results in comparison to an existing technique. |
Description: | Copyright belongs to proceeding publisher |
URI: | http://hdl.handle.net/2080/2476 |
Appears in Collections: | Conference Papers |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
2016-NCC_JBhatia.pdf | 392.02 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.