Vipula Singh et. al. / (IJCSE) International Journal on Computer Science and Engineering Vol. 02, No. 07, 2010, 2366-2374 An Algorithmic Approach for Efficient Image Compression using Neuro-Wavelet Model and Fuzzy Vector Quantization Technique Vipula Singh 1* , Navin Rajpal 2 , K. Srikanta Murthy 3 1. Electronics and communication department, Geethanjali College of Engineering Hyderabad India 2. School of Information Technology GGSIP University New Delhi, India. 3. Information science department PESIT BSK III Bangalore 560085 India. Abstract: Applications, which need to store large database and/or transmit digital images requiring high bit-rates over channels with limited bandwidth, have demanded improved image compression techniques. This paper describes practical and effective image compression system based on neuro-fuzzy model which combines the advantages of fuzzy vector quantization with neural network and wavelet transform. The emphasis here is on the usefulness of fuzzy vector quantization when it is combined with conventional image coding techniques. The implementation consists of three steps. First, the image is decomposed at different scales using wavelet transform to obtain an orthogonal wavelet representation of the image Each band can be subsequently processed in parallel. Thus, the processing speed can be much faster than otherwise. Different quantization and coding schemes are used for different sub bands based on their statistical properties. At the second step, wavelet coefficients corresponding to lowest frequency band are compressed using differential pulse code modulation. Neural network is used to extract the principal components of the higher frequency band wavelet coefficients. Finally, results of the second step are used as input to the fuzzy vector quantization algorithm. Our simulation results show encouraging results and superior reconstructed images are achieved. The effect of noise on the compression performance is also studied. Keywords: Image Compression, Fuzzy Vector Quantization, Multiresolution Analysis, Neural Network, noise. 1. Introduction Digital image presentation requires a large amount of data and its transmission over communication channels is time consuming. Numerous lossy image compression techniques have been developed in the past years[2, 5, 6, 11]. The transform based coding techniques, and in particular block- transform coding, have proved to be the most effective in obtaining large compression ratios while retaining good visual quality. Cosine transform based techniques (JPEG) have been found to obtain excellent results in many digital image compression applications [16]. Vector quantization (VQ) offers good performance when high compression rates are needed [9]. In practice, however, the existing VQ algorithms, often, suffer from a number of serious problems, e.g., long search process, codebook initialization, and getting trapped in local minima, inherent to most iterative processes. To eliminate these problems, a multi-resolution codebook is generated using fuzzy clustering techniques. These clustering techniques integrate fuzzy optimization constraints with the fuzzy- c-means algorithm [3, 13]. The resulting multi resolution codebooks generated from the wavelet decomposed images yield significant improvement in the coding process. Noise degrades the performance of any image compression algorithm [21, 22, 23]. Noise can occur during the image capture, transmission or processing and may be dependent on or independent of image content. The main objective of this paper is to present a Neuro-Fuzzy model based on wavelet transform and study the effect of noise on the image compression algorithm. The paper is organized as follows: Section 2 discusses the compression method in detail. Section 3 reports sample simulation results and section 4 provides concluding remarks. 2. Image compression model Fig 1 shows the block diagram of the image encoder. The multiresolution nature of discrete wavelet transform is a powerful tool to represent images decomposed along the vertical and horizontal directions using the pyramidal multiresolution scheme. The wavelet transform decomposes the images into a set of sub images with different resolutions corresponding to different frequency bands. 2.1 Discrete Wavelet Transform In the first step, wavelet transform is used to decompose the image into seven sub-bands. A typical 2-D DWT, used in image compression, generates the hierarchical pyramidal structure. Fig 2 shows a two-level wavelet decomposition scheme for a 512*512 digital image. This decomposition scheme produces three side bands of size 256*256 corresponding to resolution level 1 and three side bands of size 128*128, corresponding to resolution level 2. Sub-band ISSN : 0975-3397 2366