International Journal of Scientific and Research Publications, Volume 4, Issue 6, June 2014 1 ISSN 2250-3153 www.ijsrp.org Color Image Compression using Hybrid Wavelet Transform with Haar as Base Transform H. B. Kekre*, Tanuja Sarode**, Prachi Natu*** * Sr. Professor, Computer Engineering Department, MPSTME, NMIMS University ** Associate Professor, Computer Engineering Department, TSEC, Mumbai University ***Asst. Professor and Ph D. Research scholar, Computer Engineering Department, MPSTME, NMIMS University Abstract- This paper proposes color image compression method using hybrid wavelet transform and compares it with results obtained using hybrid transform and multi-resolution analysis. Haar wavelet is widely used in image compression. So here Haar transform is selected as base transform and combined with non-sinusoidal transforms like Slant, Walsh and Kekre transform. Hybrid Haar wavelet transforms is generated using Kekre’s hybrid wavelet generation algorithm. Different sizes of component transforms are used to generate hybrid wavelet transform. Haar (32x32)-Slant (8x8) gives less error as compared to Haar-Walsh and Haar-Kekre Hybrid wavelet. Performance of hybrid wavelet is compared with hybrid transform and multi-resolution by varying the size of component transforms. Mean Absolute Error (MAE) and Average Fractional Change in Pixel Value (AFCPV) are used to compare visual quality of an image. Structural Similarity Index (SSIM) of Haar-Slant hybrid wavelet is computed on 16x16 blocked images and compared with its hybrid transform and multi-resolution analysis. Least RMSE value obtained at compression ratio 32 by Haar- Slant hybrid wavelet with size 32-8 is 10.53. Mean absolute error is 7.32 in Haar-Slant hybrid wavelet with component size 16-16. Least AFCPV is 0.344 with component size 32-8 for the same. Index Terms- Haar Transform, Hybrid wavelet transform, Multi-resolution analysis, Structural Similarity Index I. INTRODUCTION Advanced technology has increased the demand of using the information in the form of images and videos. Managing this huge amount of image and video data is essential. Storage space, bandwidth requirement and transmission time are key factors that get affected due to use of this multimedia data. One way to make use of these factors effectively is reducing the amount of information being transmitted or processed. Compression of digital images plays a vital role in this context. Effective compression methods are used to obtain good quality images. An effective compression method is one which extracts characteristic features of an image and neglects redundant and irrelevant information. Transform based image coding is one of the popular image compression method. Transform when applied on images; change the image pixels to frequency domain coefficients. Desirable property of transforms is that most of the image energy is concentrated only in few significant transform coefficients. Retaining these significant coefficients and eliminating remaining coefficients results in image compression. Discrete Cosine Transform (DCT) [1] and wavelet transform are commonly used transform methods for image compression. They are used in JPEG and JPEG 2000 respectively. Recent advancements in this area show that transform based coding combined with other compression method results in better performance. In literature, transform based coding is added with vector quantization methods or neural networks or any other lossless coding method. Such methods are hybrid methods as they combine properties of two different methods. In this paper, hybrid wavelet transform based image compression has been proposed. As the name suggests, it combines characteristics of two different transforms to produce better results. II. REVIEW OF LITERATURE Wavelet based image compression has gained more popularity over DCT based image compression because of its high energy compaction property. Haar wavelets are simplest wavelets and have been widely used for compression. Haar transform is a simple, orthonormal transform proposed by Alfred Haar in 1910 [2]. It serves as a prototype for wavelet transform [3]. It allows us to encode the information according to level of details. Modified fast Haar wavelet transform (MFHWT) has been discussed by Chang P. et al. [4]. It uses one dimensional approach and FHT is used to find N/2 detail coefficients at each level for a signal of length N. Extension of this work has been proposed by Anuj Bharadwaj and Rashid Ali [5]. It works for 2D images with the addition of considering the detail coefficients for N/2 elements at each level. Haar wavelet decomposes the image into different frequency sub bands. In the technique proposed by Shridhar et al. [6] scalar quantization and DPCM is applied on image which is decomposed into different frequency sub bands. Multilevel 2-D Haar wavelet transform is used for image compression in [7] by Ch. Samson and V.U.K. Sastry. Haar wavelet using singular value decomposition has been proposed by Zunera Idrees [8]. Apart from Haar transform, 2-D wavelet