Statistical Analysis of Compression Methods for Storing Binary Image for Low-Memory Systems

Roman Slaby,Jana Nowakova,Radim Hercik

doi:10.15598/aeee.v11i4.799

Abstract

The paper is focused on the statistical comparison of the selected compression methods which are used for compression of the binary images. The aim is to asses, which of presented compression method for low-memory system requires less number of bytes of memory. For assessment of the success rates of the input image to binary image the correlation functions are used. Correlation function is one of the methods of OCR algorithm used for the digitization of printed symbols. Using of compression methods is necessary for systems based on low-power micro-controllers. The data stream saving is very important for such systems with limited memory as well as the time required for decoding the compressed data. The success rate of the selected compression algorithms is evaluated using the basic characteristics of the exploratory analysis. The searched samples represent the amount of bytes needed to compress the test images, representing alphanumeric characters.

Highlights

Digital image processing is a fast growing discipline, which is caused by the rapidly increasing development of computer technology and using of this discipline in various fields of human activity
Minimal costs of the realized systems can be achieved for example by using low-power microcontrollers, where the greatest problem can be in the insufficient size of RAM
Individual patterns represented by the binary values are compared by correlation function with the input image

Summary

Introduction

Digital image processing is a fast growing discipline, which is caused by the rapidly increasing development of computer technology and using of this discipline in various fields of human activity. The usage of systems using digital image processing in industrial applications is frequently observed trend. For accurate digitization of printed character using the correlation function it is necessary to use a large set of characters, containing all possible characters that may occur in the input image. For this reason it is required corresponding memory space for all patterns. In such systems, the patterns are stored in flash memory; the RAM is reserved for purposes of correlation functions. One of the possible solution is to use a compression algorithm to compress binary patterns representing alphanumeric digits and their subsequent storage on the mini-mum number of pages used Flash memory [1], [2], [3]

Compression Methods

Conclusion