Polygonal-feature-based shape context for flexible surface vision positioning

In the manufacturing process of curved flexible electronics, flexible electronics are transferred from a plate to a conformal curved surface before electrohydrodynamic printing. Machine vision is adopted for the localization of the process of transfer printing. However, traditional vision positioning algorithms mainly focus on the localization of a rigid object in a plate. To overcome this limitation, a novel robust and rotation-invariant algorithm, namely surface deformable localization based on shape context, is proposed for surface transfer printing of flexible electronics. First, a new localization model used in surface deformable localization is established. The proposed algorithm consists of two phases. In the off-line phase, an improved polygonal approximation is proposed to select the feature points from an image, and a rotation-invariant shape context is used to describe the feature points. In the on-line phase, the correspondence of feature points between the template image and the search image is established by the dynamic programming algorithm. Finally, the result of localization is obtained by searching for the maximum of the proposed matching mode. The experimental results show the effectiveness of the proposed method for surface deformable localization and its robustness to noise, blur, and non-linear illumination.