ANALYSIS OF WAVELET BASED COMPRESSION TECHNIQUE TO COMPRESS IMAGES USING NEURAL NETWORK Mr. Saurabh 1 Dr. Rajnesh Kumar 2 Mrs.

Transcription

1 ANALYSIS OF WAVELET BASED COMPRESSION TECHNIQUE TO COMPRESS IMAGES USING NEURAL NETWORK Mr. Saurabh 1 Dr. Rajnesh Kumar 2 Mrs. Nisha Rani 3 1 Asistant Professor & HOD, Dept. of Computer Science & Applications, OITM Hisar, Haryana (INDIA) 2 Asistant Professor, Dept. of Computer Science & Applications, CDLU Sirsa, Haryana (INDIA) 3 Asistant Professor, Dept. of Computer Science & Applications, OITM Hisar, Haryana (INDIA) 1 saurabh.charaya@gmail.com 2 rajnesh.gcnagina@gmail.com 3 charaya.nisha@gmail.com Abstract- The requirement of proficient digital information transfer requires large bandwidth. Image data compression is a very essential tool for archiving image data. The basic idea of image compression is to reduce the middle number of bits by pixel (bpp) necessary for image representation. There are various techniques available for lossy and lossless compressions. One of most popular compression techniques, JPEG uses Discrete Cosine Transformation (DCT) based compression technique. Currently wavelet based compression techniques are used for higher compression ratios with minimal loss of data. Image compression is one of the most promising subjects in image processing. Images captured need to be stored or transmitted over long distances. Raw image occupies memory and hence need to be compressed. With the demand for high quality video on mobile platforms there is a need to compress raw images and reproduce the images without any degradation. The principle behind image compression is to reduce the amount of data required for representing sampled digital images and therefore reduce the cost for storage and transmission. Image compression plays a key role in many important applications, including image database, image communications and remote sensing. We accomplish the image compression with better visibility. We work on bit area and maintain the information of the bits. Because of this as we decompress the image first the decode process is performed to get the bit information and then image restoration is applied to get back the clear visual image. We have applied the work on no. of sample images of different types. We also compared the image quality using MSE and PSNR. Image compression technique is used to reduce the number of bits required in representing image, which helps to reduce the storage space and transmission cost. Keywords- Data Compression, Image Compression, Neural Networks, Wavelet Transform, Compression Techniques. I. INTRODUCTION Data compression is the process of converting an input data stream into another data stream that has smaller size. Image compression is the application of data compression on digital images. Image compression is minimizing the size in bytes of a graphics file without degrading the quality of the image. Image Processing is a technique to enhance raw images received from cameras/sensors placed on satellites, space probes and aircrafts. Image Processing is used in various applications such as: Remote Sensing, Medical Imaging, Forensic Studies, Textiles, Material Science, Military, Film industry, Document processing, Graphic arts, Printing Industry. There are two methods available in Image Processing. ISSN Page 36

2 Analog Image Processing refers to the alteration of image through electrical means. The most common example is the television image. The Digital Image Processing generally refers to processing of a twodimensional picture by a digital computer. There are various Image Processing techniques: In Image Representation the image is considered to be a function of two real variables, for example, f(x,y) with f as the amplitude (e.g. brightness) of the image at the real coordinate position (x,y). In Image preprocessing the Scaling and Rotation of image is done. In Scaling, the work of Magnification is to have a closer view by magnifying or zooming the interested part in the imagery and with Reduction, we can bring the unmanageable size of data to a manageable limit. Rotation is used in image mosaic, image registration etc. In image enhancement, the goal is to highlight certain image features for subsequent analysis or for image display. Examples include contrast and edge enhancement, noise filtering, sharpening, and magnifying. Image analysis is concerned with making quantitative measurements from an image to produce a description of it. Image restoration refers to removal or minimization of degradations in an image. Image data compression is a very essential tool for archiving image data. There are various techniques available for lossy and lossless compressions. One of most popular compression techniques, JPEG uses Discrete Cosine Transformation (DCT) based compression technique. Currently wavelet based compression techniques are used for higher compression ratios with minimal loss of data. 1.1 Need and features of Data Compression Why present multimedia systems necessitate data compression: a) Multimedia data requires large storage space. The reduction in file size allows more images to be stored in a given amount of disk or memory space. b) Slow storage devices that do not allow playing multimedia data (mainly video data). c) The current network bandwidth does not permit real-time video data transmission. d) It also reduces the time required for images to be sent over the Internet or downloaded from Web pages. e) The main goal of image compression is to reduce irrelevance and redundancy of the image data. So, image compression is very important to reduce the storage and transmission costs while maintaining good quality. Image compression is the process of effectively coding digital images to reduce the number of bits required in representing an image. 1.2 Wavelet networks for image compression Our purpose is to use an artificial neural network and more especially a wavelet network by means of describing the network architecture specialized for the problem of image compression. This architecture includes a layer of input neurons, a hidden neuron layer and a layer of output neurons. Both of input and output layers are fully connected to the hidden layer. The feedforward propagation algorithm is used to adjust the weights of this network. 1.3 Wavelet Transform Wavelet transform (WT) of an image represents image as a sum of wavelets on multi-resolution levels. In wavelet transforms any one-dimensional function is transformed into a two-dimensional space, where it is approximated by coefficients that depend on time (determined by the translation parameter) and on scale,(determined by the dilation parameter). The zoom phenomena of the WT offer high ISSN Page 37

3 temporal localization for high frequencies while offering good frequency resolution for low frequencies. Hence, the wavelet transform is well suited to image compression. 1.4 Means to assess the Quality of Compressed Images With the growing demand for better bandwidth utilization, efficient image data compression techniques have emerged as an important factor for image data transmission and storage. Image Compression is an important area in the field of digital image processing. It deals with techniques for reducing the storage space required for saving an image or the bandwidth required for transmitting it. There are three major ways to assess the quality of compressed images: SNR, Subjective Rating (SR), and Diagnostic Accuracy (DA). Assessment of a particular compression scheme also includes compression efficiency (CE) and compression complexity (CC). For judging loss compression methods, all of the five metrics - SNR, SR, DA, CE and CC can be used. However for lossless compression, the comparison can be made solely on the basis of CE and CC. 1.5 Challenges in Image Compression Advancement of technology have produced many applications of digital imaging such as photovideotex, desktop publishing,graphics arts,colour facsimile, newspaper wire-photo transmission, medical imaging. For many other contemporary applications (such as distributed multimedia systems) rapid transmission of image is necessary. Research challenges include Developing real time compression algorithm Guaranteed quality of service in case of multimedia applications International Journal of Technical Research(IJTR) Cost minimization, because cost and time cost of transmission and storage tend to be directly proportional to the volume of data. Future video/image compression demands Improved low bit-rate compression performance Improved lossless and lossy compression Improved continuous-tone and bilevel compression Be able to compress large images Use single decompression architecture Transmission in noisy environments Robustness to bit-errors Progressive transmission by pixel accuracy and resolution Protective image security 1.6 Objective and Comparison of various image compression techniques Image compression is a result of applying data compression to the digital image. The main objective of image compression is to decrease the redundancy of the image data which helps in increasing the capacity of storage and efficient transmission. A compression scheme having a lower MSE (and a high PSNR) can be recognized as a better one. So, our objective is to achieve high PSNR with low MSE and to achieve high accuracy and generalizing ability for approximating the problems. Two of the error metrics used to compare the various image compression techniques are the mean square error (MSE) and the Peak Signal to Noise Ratio (PSNR). The MSE is the cumulative squared error between the compressed and the original image, whereas PSNR is a measure of the peak error. The mathematical formulae for the computation of MSE & PSNR are: ISSN Page 38

4 NETWORK where I(x,y) is the original image, I'(x,y) is the approximated version (which is actually the decompressed image) and M, N are the dimensions of the images, 255 is the peak signal value. The quality of image coding is typically assessed by the Peak signal-to-noise ratio (PSNR) defined as PSNR = 20 log 10 [255/sqrt (MSE)] II. NEURAL NETWORKS A neural network can be defined as a massively parallel distributed architecture for storing experimental knowledge and making it available for use. It refers to a computational paradigm in which a large number of simple computational units are interconnected to form a network, performing complex computational tasks. A neural network is a system of interconnecting neurons in a network working together to produce an output function. The output of a neural network relies on the cooperation of the individual neurons within the network to operate. Processing of information by neural networks is often done in parallel rather than in series (or sequentially). Since it relies on its member neurons collectively to perform its function, a unique property of a neural network is that it can still perform its overall function even if some of the neurons are not functioning. That is, they are very robust to error or failure (i.e., fault tolerant). There are three major learning paradigms, each corresponding to a particular abstract learning task. These are supervised learning, unsupervised learning and reinforcement learning. Neural networks can accept a vast array of input at once and process it quickly, so they are useful in image compression. III. BACK PROPAGATION NEURAL Backpropagation is a systematic method of training Multilayer Artificial Neural Networks. The Backpropagation derives from the fact that computations are passed forward from the input to the output layer. The Feed forward Backpropagation Network is a very popular model in Neural Networks. In Multilayer Feed forward Networks, the processing elements in adjacent layers are connected. This is represented by the following figure2. Some inherent features of back propagation network image data compression schemes are: (a) The network structure is massively parallel (b) The network is adaptive (c) The network examines the compressed features of the original image in a self organizing manner during the training stage (d) The intrinsic generalization property of the structure enables it to process images outside the training set (novel images) effectively. The Feed forward process involves presenting an input pattern to input layer neurons that pass the input values onto the hidden layer. The hidden layer nodes compute a weighted sum of its inputs and present the result to the output layer. IV. PROPOSED COMPRESSION ALGORITHM ISSN Page 39

5 The proposed compression algorithm is given as 1. Divide the image in image segments of some normalized size like mxm 2. Normalize the pixel value. 3. Perform the Wavelet Transformation to reduce the error. 4. Train the sub images using 3 layer back propagation Network. 5. Pass the input to the input layer, 6.From this input values some output is driven for the hidden layer and after applying some weightage the output values are derived. 7. Calculate the MSE. 8. Update weight and re perform the process. 9. Map the pixel of image with neighboring pixels 10. Calculate the compression ratio V. TOOLS USED IN MATLAB MATLAB - The Language of Technical Computing. MATLAB is a high-level language and interactive environment that enables you to perform computationally intensive tasks faster than with traditional programming languages such as C, C++, and FORTRAN. Toolbox used in MATLAB for our proposed work is Image Processing Toolbox. Image Processing Toolbox provides a comprehensive set of reference-standard algorithms and graphical tools for image processing, analysis, visualization, and algorithm development. You can perform image enhancement, image deblurring, feature detection, noise reduction, image segmentation, spatial transformations, and image registration. Many functions in the toolbox are multithreaded to take advantage of multicore and multiprocessor computers. Image Processing Toolbox supports a International Journal of Technical Research(IJTR) diverse set of image types, including high dynamic range, giga pixel resolution, ICCcompliant color, and tomography images. Graphical tools let you explore an image, examine a region of pixels, adjust the contrast, create contours or histograms, and manipulate regions of interest (ROIs). With the toolbox algorithms you can restore degraded images, detect and measure features, analyze shapes and textures, and adjust the color balance of images. VI. CONCLUSION This paper takes a detailed analysis on the most significant means of doing image compression. With the growing demand for better bandwidth utilization, efficient image data compression techniques have emerged as an important factor for image data transmission and storage. It can be observed that the convergence time for the training of back propagation neural network is very faster. Different attributes of compression such as compression ratio, peak signal to noise ratio, bits per pixel can be calculated. Image compression technique is used to reduce the number of bits required in representing image, which helps to reduce the storage space and transmission cost. We work on bit area and maintain the information of the bits. We can compare the image quality using MSE and PSNR. We can conclude that the system will provide the better visibility after the image restoration. The implementation of back propagation neural network algorithm on image compression system with good performance has been demonstrated. It has been observed that the convergence time for the training of back propagation neural network is very faster. VII. FUTURE SCOPE The field of image processing has been growing at a very fast pace. The day to day emerging technology requires more and ISSN Page 40

6 more revolution and evolution in the image processing field. Back propagation neural networks can be successfully to implement the image processing. Many experiments can also be conducted to improve the performance of the system by analyzing the experiments results with the back propagation neural network. Furthermore in future we can analyze different image coding algorithms for improvement of different parameters. VIII. REFERENCES [1] Anuj Gupta, Deepak Goel, Dr.Pankaj Gupta, et al. An Improved Adaptive Down-Sampling Approach for Low Bit-Rate Image Compression, IJMRS s International Journal of Engineering Sciences, Vol. 01, Issue 03, September [2] Rekha, Sangeeta Yogi, Geeta, et al. A Segmented Wavelet Inspired Neural Network Approach To Compress Images, International Journal of Latest Trends in Engineering and Technology (IJLTET), Vol. 1 Issue 1 May [3] Stefan Craciun, et al. Wireless Transmission of Neural Signals Using Entropy and Mutual Information Compression, IEEE Transactions On Neural Systems And Rehabilitation Engineering Ieee,2010. [4] LI Huifang,et al. A New Method of Image Compression Based On Quantum Neural Network, 2010 International Conference of Information Science and Management Engineering / IEEE,2010. [5] Aditya V. Padaki,et al. Improving Performance in Neural Network Based Pulse Compression for Binary and Polyphase Codes,12 th International Conference on Computer Modelling and Simulation / IEEE,2010 [6] Leong Kwan Li,et al. Compression of UV Spectrum with Recurrent Neural Network, / IEEE,2010 [7] Jin Wang,at al. ECG Data Compression Research Based on Wavelet Neural Network,International Conference on Computer, Mechatronics, Control and Electronic Engineering (CMCE) / IEEE,2010. [8] Luo Lincong,at al. Color Image Compression Based on Quaternion Neural Network Principal Component Analysis, / IEEE,2010 [9] Qun-ting Yang,et al. A Novel Robust Watermarking Scheme Based on Neural Network, / IEEE,2010 [10] Vilas Gaidhane,et al. Image Compression using PCA and Improved Technique with MLP Neural Network,International Conference on Advances in Recent Technologies in Communication and Computing / IEEE, [11] GUO Hui,et al. Wavelet packet and neural network basis medical image compression, / IEEE,2010. [12] Wen-Nung Lie,at al. A Perceptually Lossless Image Compression Scheme Based On Jnd Refinement By Neural Network, Fourth Pacific-Rim Symposium on Image and Video Technology / IEEE,2010. [13] T. Kohonen.et al. Self-Organizing Maps, Springer Verlag, London, 3. edition, [14] T. Martinetz and K. Schulten,et al. A neural gas learns topologies. In T. Kohonen, K. M akisara, O. Simula, and J. Kangas, editors, Artificial Neural Networks, pages North- Holland, Amsterdam, [15] J. D. Murray, W. Vanryper, and D. Russell,et al. Encyclopedia of Graphics File Formats, O Reilly UK, Cambridge, [16] W. B. Pennebaker and J. L. Mitchell,et al. JPEG: Still Image Data Compression Standard, Kluwer International, Dordrecht, [17] Erez Shermer, Neural Markovian Predictive Compression: An Algorithm for Online Lossless Data Compression, 2010 Data Compression Conference / IEEE. [18] C. Ben Amar and O. Jemai, et al. Wavelet Networks Approach for Image Compression, GVIP Special Issue on Image Compression, 2007 [19] Vijideva.R, et al. Neural Network- Wavelet Based Dicom Image Compression and Progressive Transmission, International Journal Of Engineering Science & Advanced Technology, Volume- 2, Issue-4, ISSN Page 41