A Machine Learning Approach on Image Matching in the Modern Environment

Author Name(s): S. Vidya Sagar Appaji, Prof P.V.Lakshmi, Dr P. Srinivasa Rao
Author Email: sagarsetti4u@gmail.com

Abstract

In recent trends, based on the environment, the quality of the digital photo may degrade by making it blurry, low contrast and we may get some situations where a person need to be recognized for a specific reason. To provide the quality and identity of photograph, this paper explores Image Enhancement and Image Matching algorithms using Hadoop Mapreduce Framework. The proposed algorithms can enhance the image quality and can give the approximate matching result from the dataset. Here, we need a framework that is an abstraction that provides generic functionality that is build to concentrate on image processing tasks. Map reduce paradigm framework allows the image to divide into clusters in order to facilitate distributed processing. We chose hadoop as it is known for its scalability, cost effective, flexibility, fast, security and authentication, parallel processing, and MapReduce is noted for its data processing across different computing nodes.

Introduction

Image Enhancement: The main objective is to process or alter an image so that result is more suitable than that of the original image for further image analysis or for specific application. It is the process that uses software to manipulate the stored image digitally. Filters, image editors and other tools are used for image enhancement for manipulating different properties of parts of an image or an entire image. These techniques provide more number of choices in order to improve the quality of images visually. These tools will simply change the contrast or brightness of an image or manipulate the gray scale or the red-green-blue color patterns of an image. Image enhancement includes enhancing images, sharpening and softening images, contrast change and brightening, color adjustments. Image enhancement focuses on operations of image processing which will increase the quality of image which in turn will overcome the human visual system weakness. The methods of enhancement techniques are :  Fuzzy Set Theory  Histogram Equalization  Histogram Matching  Equalized Histogram Equalization

Conclusion 

This paper presents the working of histogram equalization technique which is used for image enhancement. And further, also explains the SURF algorithm to determine similarity from the given image with the available data set of images, and this image matching is computed using feature detection , followed by feature extraction. The features obtained from the detection are matched more likely.

86 total views, 4 views today

Download PDF File

About the author: admin