Semi-automatic feature-based approach for sketch to image matching
Loading...
Date
2015-09-15
Journal Title
Journal ISSN
Volume Title
Publisher
UMT Lahore
Abstract
A semi-automatic feature-based approach for sketch-to-image matching is presented, which is used to achieve matching or recognition of sketch-based facial features with image-based facial features of human subjects. Loosely termed as “sketch-to-photo” matching, this problem is complicated by a wide variety of factors in both sketches and photos, some of which include angle, color, lighting, and scale. Our proposed approach overcomes these difficulties through normalization of inputs and selection of prominent and reliable feature points. The basis of comparison between corresponding feature points is the ratio of Euclidean distances, which makes the approach scale-independent. The approach promises discernible improvements in accuracy and efficiency.