Kajian Peningkatan Kualitas Ekstraksi Fitur Berdasarkan Pola Gerakan Mata Untuk Kepentingan Rekognisi

Abstract

This research examines how to recognize objects in digital images with results that can be explained logically like the perspective of the human eye. Pattern recognition techniques using statistical methods cannot provide a logical description of the recognized objects, because this concept considers the features used as probability classes, so the essence of the way the human eye sees cannot be demonstrated. The syntactic method is also not able to provide a logical description of the object that is recognized according to the perspective of the human eye, this concept prioritizes low-level features for the recognition process. So, in this research, we examine several syntactic and statistical recognition methods that adapt some of the standard abilities of the human eye. Features such as lines, chain codes, and colors have been able to define objects in images, and approach human reasoning. Simple Human Eye Movement Analysis, can help us to detect the relationship between line, true color, and chain code to show the object unity. We hope that developing this approach will enrich the object pattern recognition method to be simpler and faster.