Machine Learning-Based Scene Classification Using Thepade’s SBTC, LBP, and GLCM

Abstract

AbstractPhotographs have become a necessary part of human life. Images are being captured all across the world. One of the significant issues in scene classification is that the photographs are taken in a variety of locations. There are many diverse scenes in these locations, such as an airport, a restaurant, a cafeteria, and a forest. Indoor scene classification and outdoor scene classification are the most common types of scene categorization. Scene classification is challenging, which poses issues in research and image processing. Image classification becomes challenging because of the numerous distinct items present in various sceneries, causing a machine to become confused about the scene’s categories. The current challenge with scene classification results in low-classification accuracy. The object comes in various shapes and sizes and can be found in various locations, which is one of the reasons for the scene classification’s poor accuracy. We will perform feature extraction for pattern recognition and multiple machine learning models for image categorization using the intel image classification dataset. Thepade’s SBTC n-ary feature extraction approach for RGB and LUV color plane models and fusion LBP and GLCM is presented in this study. When the three individual feature extraction methods are compared with the fusion of Thepade’s SBTC and GLCM with RGB color plane, the fusion of Thepade’s SBTC and GLCM with RGB color plane yielded greater accuracy.KeywordsScene classificationFeature extractionThepade’s SBTCLocal binary patternGLCM