Oyindamola Oduba made a poster presentation at 2018 TTU Undergraduate Research Conference

Oyinda presented a poster on March 27, 2018 at Texas Tech University, Undergraduate Research Conference 2018.

"Feature Identification of Sketches Based on Spatial Organization Principles"

Abstract

In many applications, features of images or graphical representations can be represented in different scales, from low-level pixel based information to high-level structures, such as parts of, or whole objects in an image. Analogously, features of sketches can be defined as groups of geometric elements resulting in composite units that can be shown to have a functional aspect. Based on this concept, this paper proposes a systematic qualitative procedure that can be used to identify the features of a sketch. The key characteristics of the procedure is its reliance on the spatial relationship of the visual elements composing the sketch by combining concepts from both hierarchy-of-parts and constellation-of-parts approaches, as well as the Gestalt principles of visual organization. The proposed feature identification procedure was based on two main concepts: (i) division of the sketches into atomistic geometric elements and, (ii) grouping of the atomistic elements according to their spatial arrangement into features using hierarchy-of-parts, constellation-of-parts, and Gestalt principles. The results obtained for the three different sketches showed that the proper identification of the visually largest and most centrally positioned element, or group of elements, in a sketch was related to the easiest grouping and identification of the remaining features. It has also been shown that the spatial arrangement of the elements of the sketch allowed them to be grouped into features in an intuitive and logical manner, eliminating any necessity of prior knowledge on the purpose and functionality of the represented artifact to identify the features of a sketch.