Analysis Of Sift And Surf Algorithms For Image Mosaicing On Embedded Platform

Ooi , Chong Wei (2015) Analysis Of Sift And Surf Algorithms For Image Mosaicing On Embedded Platform. Masters thesis, Universiti Sains Malaysia.

PDF - Submitted Version
Download (489kB) | Preview


In computer vision, image mosaicing or stitching is a common active research area. Image stitching process is a process of compositing images which contain similar scene into a larger image. The union of these input images is called panoramic image. Image stitching techniques are classified into two types. First technique is direct technique whereas another is known as feature-based technique. Significant pros of feature-based method are in terms of robustness and speed. As a result, panoramic image is created faster and contains quality improved. In this paper, a real time on board image mosaicing system based on SURF feature based techniques is proposed. Performance comparison between SURF and SIFT is made. To obtain matching point between images, Flann Based Matcher is used. Next homography estimation is performed by using RANSAC algorithm. Perspective transform is applied to obtain a transformation for mapping a two dimensional quadrilateral into another. Lastly, images are warped and composited into single scene. Experimental results shows that SURF and SIFT are robust algorithm performing stable key point detection. These techniques are invariant to scale and rotation. SURF technique has better performance with respect to speed. Implementation and experimental are done in Raspberry Pi board with built-in 512MB RAM and 700MHz processor.

Item Type: Thesis (Masters)
Additional Information: Accession No:875005909
Subjects: T Technology > TK Electrical Engineering. Electronics. Nuclear Engineering > TK7800-8360 Electronics
Divisions: Kampus Kejuruteraan (Engineering Campus) > Pusat Pengajian Kejuruteraan Awam (School of Civil Engineering) > Thesis
Depositing User: Mr Mohd Fadli Abd Rahman
Date Deposited: 24 Aug 2018 02:18
Last Modified: 24 Aug 2018 02:18

Actions (login required)

View Item View Item