文章基本信息

标题：Performance Comparison of Hybrid Haar Wavelet Transform with Various Local Transforms in Image Compression using Different Error Metrics
本地全文：下载
作者：Dr. H. B. Kekre ; Associate Professor Tanuja Sarode ; Miss Prachi Natu 等
期刊名称：International Journal of Image Processing (IJIP)
电子版ISSN：1985-2304
出版年度：2014
卷号：8
期号：4
页码：186-203
出版社：Computer Science Journals
摘要：A novel image compression using hybrid Haar wavelet transform has been proposed in this paper. Hybrid wavelet transform is generated using two different orthogonal transforms. Haar transform acts as a base transform and other sinusoidal transforms like DCT, DST, Hartley and Real-DFT are paired with Haar transform to generate hybrid Haar wavelet. Among these four pairs Haar-DCT hybrid wavelet gives lower error as compared to Haar-DST, Haar-Hartley and Haar-Real-DFT. Performance of Haar-DCT hybrid wavelet is further analyzed using multi resolution hybrid wavelet and Haar-DCT hybrid transform. Experimental results show that hybrid wavelet with component size 16-16 gives lower error at higher compression ratios than multi resolution analysis and hybrid transform performance. Performance is measured using RMSE which is traditional parameter to measure error. Lowest RMSE obtained is 9.77 at compression ratio 32 using Haar-DCT Hybrid Wavelet with component size 16-16. Various other error metrics like MAE, AFCPV and SSIM are used to measure error. Lowest MAE and AFCPV are observed at compression ratio 32 are in Haar (16x16) -DCT (16x16) hybrid wavelet having values 6.86 and 0.31 respectively. When blocked SSIM is applied on 16-16 Haar-DCT hybrid wavelet it gives value 0.993 at compression ratio 32 which is closer to one indicating that good quality of compressed image is obtained.
关键词：Hybrid Transform; Haar Wavelet; SSIM; MAE; Image Compression.