IBTC-DWT Hybrid Coding of Digital Images Authors: Ali Abdulhafid Elrowayati, Zakaria Suliman Zubi, Electronic Engineering Department, Computer Science Department, The College of Industrial Technology, Faculty of Science Misurata, Libya Sirte University Sirte, Libya02/07/13 1
Abstract1. A hybrid IBTC-DWT encoding combines the simple computation and edge preservation prosperities of interpolative block truncation coding(IBTC) and high compression ratio of discrete wavelet transform(DWT).2. This implemented yields significantly lower coding delay than DWT alone, and to achieve a reduced bit rate is also proposed and investigated.3. In this hybrid IBTC-DWT algorithm, the resulting high- means and low-means sub images from IBTC algorithm are coded using DWT transform.4. Simulation results showed that good performance was demonstrated in terms of compression ratio, bit rate and reconstruction quality.02/07/13 3
Introduction► Image Coding Compression of digital images has been a topic of research for many years and a number of image compression standards has been created for different applications. The role of compression is to reduce bandwidth requirements for transmission and memory requirements for storage of all forms of data.02/07/13 5
Introduction► There are two main families for image compression: Lossless image compression techniques ► Lossless have the disadvantage of being limited in term of compression rate. Lossy techniques ► Lossy techniques allow larger compression rates. ► while introducing some distortion in reconstructed images. ► In order to improve compression rates, we are interested in the second family of techniques. ► In this paper, we propose a novel method of encoding an image using both the interpolative block truncation coding (IBTC) and discrete wavelet transform (DWT) to achieve significant improvement in digital image compression performance.02/07/13 6
Introduction► Block Truncation Coding (BTC) BTC is a block-based lossy image compression First developed in 1979 for grey scale image coding The output data of BTC for an image block contains one bitmap and two quantization levels BTC has very few computations, edge- preserving ability; but only a medium02/07/13 compression ratio. 7
An Example of BTC Encodingw 4×4 image pixels Bitmap 140 142 144 145 0 0 1 1 146 141 146 142 1 0 1 0 145 141 144 142 1 0 1 0 142 138 141 144 0 0 0 1 Mean value X=142.5 # of 0 is 9Original Image Variance value ρ=2.199 # of 1 is 7, q=7 q 7 X L = X −σ = 142.5 − 2.199 * = 141 m−q 9 Two quantization levels m−q 9 XH = X +σ = 142.5 + 2.199 * = 145 q 7 8
An Example of BTC Decoding Bitmap 0 0 1 1 141 141 145 145 1 0 1 0 145 141 145 141 Decoding 1 0 1 0 145 141 145 141 0 0 0 1 X L , bi = 0, 141 141 141 145 oi = ˆ X H , bi = 1.X L = 141 X H = 145 Reconstructed pixels 9
Introduction► INTERPOLATIVE Block Truncation Coding (IBTC) IBTC algorithms are based on the fact of the adjacent image pixels have high degree of correlation and the resulting bit-maps will also high degree of correlation. Only half of the bits of bit maps for each block are transmitted or stored and the other are interpolated. IBTC uses only 8 bits of 4× 4 bit-maps instead of 16 bits, thereby reducing the bit rate from 2 bits/pixel to 1.5 bits/pixel.02/07/13 10
Introduction► Discrete wavelet transform (DWT) DWT can be efficiently used in image coding applications because of its data reduction capabilities. Unlike the case of Discrete Cosine Transform (DCT) which based on cosine functions, DWT has some properties, making it a better choice for image compression than DCT, especially for image on higher resolutions. DWT coding gives better representation of bits with localization in both the spatial and frequency domains02/07/13 11
Introduction► The main idea of the proposed method: The presented hybrid IBTC-DWT algorithm combines the simple computation and edge preservation prosperities of IBTC and high compression ratio of DWT02/07/13 12
The proposed scheme02/07/13 14 Fig 1 System diagram
The proposed scheme► For a 512×512 input images with 4×4 blocks► the sub sampled images are 128×128 in size. The sub sampled-images have details and features which must be preserved, since any distortion involved here will be distorted over all of the pixel in each reconstructed IBTC block. DWT is directly implemented on both the high-mean sub image and the low-mean sub image. For example, when using level-2 of decomposition, and take the important coefficients with high energy. Since the size of sub images is relatively small (16 times less than original image) the computational complexity is02/07/13 15 reduced.
Simulation results► Test image : Lenna ,size is 512×512 bit resolution is 8 bit► DWT transform has been used with different scalar quantization, where the significance of coefficients are directly related to its magnitude as well as their sub bands after wavelet decomposition at different low bit rates. (0.82 bpp as example)02/07/13 17
Simulation results► In previous table show that the proposed algorithm give better performance in terms of The MSE and PSNR compared to the result of IBTC-DCT algorithm in .► In Fig.1, shows one of the test images and its reconstructed version using the proposed algorithm at different low bit rates.02/07/13 19
Simulation results Fig.1. The original and reconstructed image using different bitrate. (a) original Lena. (b) Reconstructed using 0.102bbp (c) Reconstructed using 0.50bbp (d) Reconstructed using 0.820bbp.
Conclusion► Digital Image Compression has been achieved using the proposed IBTC-DWT algorithm.► Comparison between the numerical results obtained by proposed algorithm with the corresponding ones obtained by of IBTC-DCT algorithm in .► This IBTC-DWT algorithm gives good quality reconstructed images at low bit rate.02/07/13 22