A PSO-ANLBF based automated feature extraction method

Abstract

Image is a much better way of explanation than documentation. The summary of any problem area can be easily described with images. On the other hand, documentation takes much time to make people understood the real problem. With the developing technology, images have been so important in many sectors of the modern world. The medical reports, security verification, boundary measurement- these types of services need very regulable images. Thus image processing is drawing an important role in modern technology. Filtering and feature extraction are the major steps of image processing. There are various types of methods for feature extraction and filtering. Non-local filtering and bilateral filtering are two efficient methods of filtering. We come up with an idea of a hybrid approach of filtering using both non-local and bilateral filtering, called ANLBF (adaptive nonlocal bilateral filtering). The filtering system will efficiently contribute in noise reduction and color enhancement and the output image will be converted into binary image and later will be classified utilizing threshold segmentation. The feature extraction is based on particle swarm optimization technique that helps to reduce the feature vectors. The performance of the proposed approach is acquired using the classification accuracy rate that shows that the approach is effective with minimum 82% accuracy and maximum 91% accuracy rate.