Effect of Features and Angle on Gray Level Co-occurrence Matrix Feature Extraction on Accuracy for Object Classification

Abstract

The cars that can move on their own or have the ability to drive without assistance from humans are called autonomous cars. The development of various types of driverless vehicles is currently underway. Where in the future the computer system will replace the role of humans in driving vehicles. However, the problem in autonomous cars that deserves attention is the need for high security. Early warning systems are needed in autonomous car systems to detect objects in front of them. This is necessary to avoid accidents, especially when on the highway. In this study, researchers designed a system for vision-based vehicle detection in detecting cars in front of them. The detection algorithm used has two main components, namely color feature extraction using GLCM values, and 6 parameter testing of GLCM dissimilarity, correlation, homogeneity, contrast, ASM and energy. In this study using the SVM (Support Vector Machine) algorithm for the classification algorithm. Good accuracy results are found in the ASM feature and using an angle of 450, which is 88%.