Image Compression Data Mining

This system has been created to perform improved compression using Data Mining Algorithms.

Running Instructions:

1. Jepeg_Haufmann.m - > This performs the jpeg compression
2. testf2.m -> This performs the pattern mining and huffman encoding
3. decode.m -> This performs the decoding
4. combine.m -> This combines all the files
5. measures.m -> This provides the metrics

Algorithms used:

1. JPEG Compression
2. ARM - Apriori Algorithm
3. FP Huffmann Encoding and Decoding

Software: MATLAB 2014a

Steps:

1. Image converted to matrix
2. Algorithms run over dataset
3. Output image regained from matrix

Efficiency = 85.6 % with reference to exisiting system Ideal image size:

1. 8X8
2. 16X16
3. 32X32
4. 64X64
5. 128X128
6. 256X256
7. 512X512

Not ideal for Large sized images

Citation:

If you use this code in your work then please cite the paper - Lossy Image Compression using Frequent Pattern Mining based Huffman Encoding with the following:

@INPROCEEDINGS{8487850, 
author={S. {Biswas} and N. {Chennu} and H. {Valveti} and C. {Oswald} and B. {Sivaselvan}}, 
booktitle={2017 14th IEEE India Council International Conference (INDICON)}, 
title={Lossy Image Compression using Frequent Pattern Mining based Huffman Encoding}, 
year={2017}, 
volume={}, 
number={}, 
pages={1-6}, 
keywords={data compression;data mining;discrete cosine transforms;Huffman codes;image coding;runlength codes;frequent pattern mining;lossy compression techniques;lossless compression techniques;lossy algorithms;lossless algorithms;modified DCT algorithm;decoding;modified run-length encoding;compression ratio;image quality;lossy image compression;Huffman encoding;Image coding;Transform coding;Matrix converters;Discrete cosine transforms;Quantization (signal);Data mining;Itemsets;Lossy algorithm;Modified DCT algorithm;Compression ratio;Frequent Pattern Mining;Huffman Encoding;PSNR}, 
doi={10.1109/INDICON.2017.8487850}, 
ISSN={2325-9418}, 
month={Dec},}