FUNCTIONAL SEGMENTATION OF STROKES FOR PRODUCT SKETCH UNDERSTANDING

Abstract

The paper introduces the concept of functional segmentation for segregating the strokes in a product concept sketch into functionally meaningful groups. Analysis of the distributions of starting points, time and mean pressure associated with the strokes in sketches revealed that shape defining and shape detailing strokes are separable through thresholding of the stroke parameters. The inter-stroke time plot exhibited characteristic spikes which enabled stroke-sets to be conveniently segmented in to temporal groups. The temporal evolution of shape defining elements and the genesis of feature hierarchy showed significant correlation. This algorithmic stroke organization showed significant match with the segmentation done by designers interactively. Spatial grouping algorithms developed using Gestalt principles isolated shapes in a sketch well. A unified multi-parameter grouping and averaging technique is also presented for obtaining single stroke sketches without losing the feature structure. The work thus demonstrates that algorithmic functional segmentation require both geometric and non-geometric data (viz. coordinates, time-stamp and pressure associated with the strokes) and that the role based classification of strokes helps to relate functional composition of the product and temporal evolution of the sketch. The work identifies how the designers emotion, rational thinking and visualization skill influence the product concept sketching process.