HAPTIC VISUALIZATION USING VISUAL TEXTURE INFORMATION

Haptic enables users to interact and manipulate virtual objects. Although haptic
research has influenced many areas yet the inclusion of computer haptic into
computer vision, especially content based image retrieval (CBIR), is still few and
limited. The purpose of this research is to design and validate a haptic texture search
framework that will allow texture retrieval to be done not just visually but also
haptically. Hence, this research is addressing the gap between the computer haptic and
CBIR fields.
In this research, the focus is on cloth textures. The design of the proposed
framework involves haptic texture rendering algorithm and query algorithm. The
proposed framework integrates computer haptic and content based image retrieval
(CBIR) where haptic texture rendering is performed based on extracted cloth data. For
the query purposes, the data are characterized and the texture similarity is calculated.
Wavelet decomposition is utilized to extract data information from texture data. In
searching process, the data are retrieved based on data distribution.
The experiments to validate the framework have shown that haptic texture
rendering can be performed by employing techniques that involve either a simple
waveform or visual texture information. While rendering process was performed
instability forces were generated during the rendering process was due to the
limitation of the device. In the query process, accuracy is determined by the number
of feature vector elements, data extraction, and similarity measurement algorithm. A
user testing to validate the framework shows that users’ perception of haptic feedback
differs depending on the different type of rendering algorithm. A simple rendering
algorithm, i.e. sine wave, produces a more stable force feedback, yet lacks surface
details compared to the visual texture information approach.