DEVELOPMENT OF POINT CLOUD DESCRIPTORS FOR ROBUST 3D RECOGNITION

• Main Author: OSMAN ALI, YASIR SALIH
• Format: Thesis
• Language: English
• Published: 2015
• Subjects: TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://utpedia.utp.edu.my/21543/1/2015%20-ELECTRICAL%20%26%20ELECTRONIC%20-%20DEVELOPMENT%20OF%20POINT%20CLOUD%20DESCRIPTORS%20FOR%20ROBUST%203D%20RECOGNITION%20-%20YASIR%20SALIH%20OSMAN%20ALI.pdf; http://utpedia.utp.edu.my/21543/

Object recognition allows machines to understand the nature of objects it encounters in the surrounding environment. Descriptors are the most important element for robust object recognition system as they assign a unique identification to each object that withstands pose and illumination variations. Among many surface descriptors proposed in the literature, orientation based descriptors are the most commonly used, however they are represented with large and sparse histograms. Moreover, existing descriptors are highly susceptible to noise and viewpoint variations. These limitation prevented object recognition algorithms from being implemented on embedded devices (e.g: smartphones) and mobile robots specially for mapping applications.