ABSTRACT
This chapter provides an improvement of replacing the Harris-affine detection method of G. Csurka et al. by a random sampling procedure together with an increased number of sample points. It describes feature extraction and description and discusses object categorization and presents the improved method based on random sampling. Image categorization aims to label or classify images into one of the predefined categories. It attempts to retrieve all the images from the same category as a given query image. The attributes of similarity vary from system to system, which are mostly based on color, texture, and shape features. The bag-of-feature and feature vocabulary-based approaches have been presented for image categorization due to their simplicity and competitive performance. Feature extraction is the method for locating points of interest in an image that can be added to a database to be searched later in order to identify objects.
