Shape context with bilinear interpolation

Shape context is intended to be a way of describing shapes that allows for measuring shape similarity and recovering point correspondences. It is widely used in object recognition, and gets encouraging performance. But shape context is easily affected by noise points and slightly transition. In image processing, bilinear interpolation is practical and effective, allowing the resulting image appears smoother rather than jagged rendering, and more robust to slightly transition. In this paper, we use the bilinear interpolation to calculate shape context. Experimental result on 20 classes of object containing 500 images shows that robust objects matching can be achieved with our proposed method.