Abstract: Sketch-based communication is nonewer; it is the oldest form of writing. In which sketch shows rough orunfinished drawing of object. The photorealistic image shows attention torealistic detail. There is a large appearance gap in user sketches andphotorealistic images therefore to bridge this gap is fundamental challenge inSBIR. The existence of noisy edges on photo realistic image degrades retrievalperformance and to bridge this gap there is framework consisting of linesegment descriptor and noise impact reduction algorithm. Proposed descriptorextracts edges and captures the relationship between those edges. Objectboundary selection algorithm used to reduce the noisy edges for which thehypothesis is used to maximize retrieval score, for which multiple hypothesesare generated.
Keywords: descriptor, sketch retrieval, edge based, histogram, line relationship.