Abstract: A novel freehand sketch retrieval method was presented, which is based on affine transformation adaptive shape and skeleton features. Firstly, a Fourier-based approach to dimensionality reduction and smoothing was employed to process object contour data. And, after normalization, the sampled contour data was used to extract contours and regional features of objects. Then, a new affine adaptive skeletonization algorithm was presented to extract the skeletons of objects and a new skeleton tree descriptor and matching algorithm were derived. At last, a new sketch retrieval method combining contour region and skeleton features was presented. The experimental results show that the method is robust to object affine transformation (translation, scaling, rotation) and noise corruptions.

Key words: contour, region, skeleton, affine transform, sketch retrieval