Huffman Based LZW Lossless Image Compression Using Retinex Algorithm

Dalvir Kaur.
International Journal of Advanced Research in Computer and Communication Engineering Vol. 2, Issue 8, August 2013

Review oleh: Mega Satya C / 136060300111018

Sistem Komunikasi dan Informasi Fakultas Teknik Elektro Universitas Brawijaya Malang 2013

Referensi
Judul Jurnal:
Huffman Based LZW Lossless Image Compression Using Retinex Algorithm. Penulis: Dalvir Kaur
(Master of Technology in Computer Science & Engineering, Sri Guru Granth Sahib World University, Fatehgarh Sahib, Punjab, IndiaMaster of Technology in Computer Science & Engineering, Sri Guru Granth Sahib World University, Fatehgarh Sahib, Punjab, India)

Abstraksi

Image Compression Encodes original image with few bits Reduce irrelevance and redundancy of the image data Can reduce the transmit time over the network and increase the speed of transmission

Abstraksi

Lossless Compression and Propose Technique

No data loss 1. Compress image with Huffman coding 2. LZW Coding and Decoding 3. Retinex Algorithm to enhance the contrast of image and improve the quality of image To increase the Compression Ratio (CR), Peak Signal of Noise Ratio (PSNR), and Mean Square Error (MSE) Use Matlab Software

Introduction

Image Compression
Means: the reduction of the size of image data, while retraining necessary information Mathematically: transforming a 2D pixel array into a statically uncorrelated data set So, image compression is used to minimize the amount of memory needed to represent an image

Introduction

Block Diagram of Image Compression System

Introduction

Fundamental Image Compression

Neighbouring pixels of image are correlated and therefore contain redundant information
Redundancies Reduction
Removing duplication from signal source (image/video)

Fundamental components of Image Compression

Irrelevancy Reduction
Removing duplication from signal source (image/video)

Introduction

Three types of image redundancies

Coding Redundancy : fewer bits to represent frequently occuring symbols Interpixel Redundancy: neigbouring pixels have almost same value Psycho visual redundancy: Human visual system cannot simultaneously distinguish all colors

Introduction

Need of Image Compression

To create faster loading web pages (for Web Designers) Save lof of unnecessary bandwidth Attach photos to email , send the email more quickly and save bandwidth cost Store more images in hard disk saving memory space
Suppose we need to download a digitized color photograph over a computer's 33.6 kbps modem. If the image is not compressed (a TIFF file, for example), it will contain about 600 kilo bytes of data.

Introduction

Image Compression Coding

is to store the image into bit-stream as compact as possible and to display the decoded image in the monitor as exact as possible
The basic flow of image compression coding

Introduction

The basic flow of image compression coding

The encoder receives the original images Image file will be converted into a series of binary data (bit stream) The decoder receives the encoded bit stream The decoder decodes it to form decoded image The total data quantity of decoder image fewer than the original image image compression

Introduction

Types of Compression
Lossy coding techniques the reconstructed image contains degradation relative to the original image
Transformation coding, Vector Quantization, Fractal Coding, Block Truncation coding, Sub Band Coding

Lossless coding techniques the reconstructed image is numerically indentical to the original image
Run Length Enoding, Huffman Encoding, LZW Coding, Area Coding

Proposed Work

Algorithm
Step 1 Read the image on to the workspace of Matlab
Step 2 Call a function to find the symbols (pixal value) Step 3 Call a function to calculate the probability of each symbol Step 4 Arrange probability symbols in decreasing order. Lower probabilities are merged. Continue this step until only two probabilities are left. Code are assigned According to rule that the highest probable symbol will have a shorter length code

Step 5 Mapping of the code words to the corresponding symbols. Result: Huffman codewords

Proposed Work

Algorithm (continued)
Step 6 Concatenate all the Huffman code words and apply LZW encoding. Result: LZW dictionary and final encoded values (compressed data)

Step 7 Apply LZW decoding process on final encoded values and output the Huffman code words
Step 8 Apply Huffman encode value on LZW encoding process

Step 9 In final apply the Multiscale Retinex Algorithm on compressed image to enhance The quality and color of the image
Step 10 In last step the recovered image is generated

Proposed Work

Flowchart of Huffman based LZW Lossless Image Compression sing Retinex Algorithm

Proposed Work

Huffman Coding and Decoding Process Use the probability distribution of the alphabet of the source to develop the code words for symbols.

Proposed Work Flowchart: Huffman Algorithm

Proposed Work

LZW Coding and Decoding The algorithm : scan a file and search the sequence of data or string that occur more than once in a file LZW replacing strings of characters with single codes without doing any analysis of the incoming data

Proposed Work

LZW Encoding

Proposed Work

LZW Decoding

Proposed Work

Image Enhancement using Retinex Algorithm

Retinex : Retina and Cortex abbreviate Retinex balance three aspects in compress the dynamic range of grayscale, edge enhancement, and color constancy Basic principles: brightness image and reflection image

Proposed Work

Retinex Provides:
Great dynamic compression Increased sharpness and color Fast algorithm that is used with different color bands

Retinex:
Single-scale Retinex (SSR) Multiscale Retinex (MSR) Multiscale Retinex for Color Restoration (MSRCR)

Proposed Work

Multiscale Retinex Algorithm

N = number of scales Rni = the ith component of the nth scale Wn = weight values

Result

The Graphical User Interface

Result
Lena Image

House Image

Result

Conclusion

Different algorithms have been evaluated in terms of the amount of compression they provide, algorithm efficiency, and susceptibility to error Reproduced image and the original image are equal in quality by using Retinex Algorithm, as it enhances the image contrast using MSR

Future Scope

Improvement of compression ratio using the new techniques The proposed technique can be experimented on different kinds of data sets (audio, video, text) as till now it is restricted to image New methods can be combined and proposed that decreases the time complexity incurred in creating dictionary in LZW algorithm Larger dataset could be a subject for future research

Reviewer Suggestions
Sebaiknya dijelaskan tipe data citra yang digunakan untuk percobaan, serta ukuran citra asli dan hasil kompresi. Dari hasil pengujian, sebaiknya dijelaskan arti pencapaian nilai rasio kompresi, MSE, dan PSNR yang dihasilkan dari penerapan metode kompresi.

End..