Investigating image processing algorithms for provision of information in rock art sites using mobile devices

• Main Author: Olojede, Ayodeji
• Other Authors: Suleman, Hussein
• Format: Dissertation
• Language: English
• Published: University of Cape Town 2017
• Subjects: Computer Science
• Online Access: http://hdl.handle.net/11427/23783

The term cultural heritage spaces incorporates places, objects and practices of cultural and historical significance. Examples include the Southern African rock art heritage sites. Rock art is an archaeological term used to describe man-made markings on stones. Studies have revealed that visitors to rock art sites usually do not understand the meaning of the rock art artefacts they are looking at due to a lack of descriptive information necessary to frame the artefact in the proper cultural and historical context. Instead, rock art sites offer humans as tour guides. One problem observed with human tour guides is that they often do not provide enough information about the artefact and they also do not answer questions to the satisfaction of most visitors. Also, human guides are a limited and expensive resources and do not always provide a personalized experience for each visitor. Therefore, in this research, an alternate interpretation mechanism that gives visitors a personalized interaction with rock art artefacts is proposed. We introduce Heritage Vision, a mobile guide application that enables visitors to take a picture of a rock art artefact of interest and automatically presents information about the artefact to the visitor. This is done via a content based image retrieval system with the aid of image processing. We investigate 3 image processing algorithms for digital recognition of rock art images on mobile devices. The ubiquitous nature and recent technological advances has made mobile devices the preferred medium. Image processing algorithms such as Scale- Invariant Feature Transform (SIFT), Speeded-Up Robust Features (SURF) and Oriented Fast and Rotational Brief (ORB) have been incorporated in a mobile guide prototype and their performance has been evaluated. Performance evaluation has revealed that the ORB algorithm has a better and acceptable performance over the SIFT and SURF algorithms. A user experiment was performed to evaluate the usability of the application prototype using SUMI (software usability measurement inventory) and the result obtained shows a SUMI global scale (perceived quality of use) score of above average, suggesting that such a solution is feasible.