Feature-Based Image Patch Approximation for Lung Tissue Classification Using RGLBP and MCHOG

Abstract

Since there are over 150 different types of lung tissue disorders that can occur, the main goal of this project is to identify abnormal lung tissue. The anomaly of the lung tissue was first identified in order to diminish this condition, and the proper treatment was then administered through clinical practise. As a result, the imaging system is used to detect the issue utilising some of the currently available approaches. The methodology utilised in this suggested method to extract the feature from the collected HRCT image uses the rotation-invariant Gabor-local binary patterns (RGLBP) texture descriptor with multi-coordinate histogram of oriented gradients (MCHOG). The dataset image is taken for the detection of the irregularity in order to get better results and images of higher quality. The picture is reduced to include only the various lung tissue locations necessary for the SVM classifier to identify the state of abnormality. In this procedure, the test features and train features are classified using the SVM classifier to determine the condition of the anomaly. The test feature and the train feature play a vital part in the classification procedure for the classification purposes. The train feature is only the feature that was got from the train picture, whereas the test feature is only the feature that was gained from the test image.