Indian Journal of Science and Technology
Year: 2016, Volume: 9, Issue: 29, Pages: 1-12
S. S. Tamboli1 * and V. R. Udupi2
1 Department of Electronics and Telecommunication Engineering, [email protected]
2 Department of Electronics and Communication Engineering, [email protected]
*Author for correspondence
Department of Electronics and Telecommunication Engineering,
Email: [email protected]
Objectives: The basic aim of this paper is to develop a coding method which will give effective compression by retaining image accuracy with lower computational overhead in image coding. The objective for obtaining such a coding technique is made by the development of an improvised preprocessing approach followed by a modified planar coding with the transformation made from wavelet to multi-wavelet. Methods/Statistical Analysis: This paper analyzes the performance of existing wavelet based compression techniques. Also develops a coding approach so as to obtain effective compression by retaining image accuracy with lower computational overhead in image coding. The overall implementation consists of improvised preprocessing followed by a modified planar coding with the transformation made from wavelet to multiwavelet approach. The main problems focused in this research work are 1. Preprocessing (Filtering), 2. Representing image coding coefficients and 3. Coding schemes for multi-Bit rate compatibility with minimum representation. For noise removal a modified weighted filtration approach is proposed. With the proposed approach an improvement in coding efficiency is achieved. The simulation observation evaluates the proposed approach and the comparative analysis of the proposed approach presents the improvements achieved. The performance of proposed selective MWVLT (S-MWLT) is compared with the conventional Multi Wavelet coding (MWVLT) and DWT based coding. The assessment is carried out by observations on various test samples. Findings: To test the operation performance for developed system the PSNR, RMSE for the system is evaluated under different medium distortion level. The observation illustrates that the obtained visualization of the filtered result using weighted filtration is comparatively more accurate than the conventional filtration approach while RMSE value for the proposed approach is decreased to about 40 units as comparative to the conventional approach. It is observed that the obtained filtration is improved with the block size increment. At N=4 the obtained filtration is comparatively accurate. Analysis of different wavelet transformations at variable bits per pixel is carried out. From analysis it is clear that Mean Square Error (MSE) and computational time is less in symplet transformation as well as high PSNR is obtained for symplet transformation. Analysis of DWT, MWVLT, S-MWVLT at different noise variance is carried out. The MSE value is observed minimum for the proposed S-MWLT coding than DWT and MWLT. This is due to minimal correlative band selection, the MSE of recovered sample is observed to be lower. While the compression achieved for the proposed S-MWVLT coding, is comparatively higher than the DWT based coding. The overhead is lower in case of DWT based, however in comparison to MWVLT coding; proposed S-MWVLT coding has lower overhead. The PSNR for the proposed approach is observed to be improved by a factor of 7 dB in comparison to MWVLT coding and about 10 dB in comparison to DWT based coding. Also at higher coding rate proposed S-MWVLT coding performs better than DWT, MWVLT. Application/Improvements: The image compression using selective Multiwavelet coding can be extended for video compression and can be applied in multimedia communication. It can also be extended for other image processing applications (may be face recognition, passport size image compression, etc.) for better results. This technique can also be combined with image security.
Keywords: Adaptive Filtering, Compression Ratio (CR), DWT, Hybrid Hierarchical Coding, Image Compression, Multiwavelet Transformation, MWLT, Peak Signal to Noise Ratio (PSNR), Root Mean Square Error (RMSE), Selective Coding
Subscribe now for latest articles and news.