Voting-based decision framework for optimum selection of interpolation technique for 3D rendering applications

This paper investigates a novel decision framework for efficient selection of interpolation curve based on distance minimization for 3D rendering applications. The point clouds obtained from low resolution 3D scanners like Microsoft's Kinect or from sparse reconstruction algorithms usually fail to provide accurate information about the surface, either due to occlusions during the scanning process or inability of the scanner to generate a dense model of the surface. The proposed decision framework selects the best interpolation technique on a local basis utilizing the voting parameters obtained from the original point cloud. This framework enables us to obtain the comparatively best fit interpolation curve for upsampling due to the decisive feature of the framework. Experimental results are carried out using two interpolation techniques viz., quadratic spline interpolation and cubic spline interpolation technique to demonstrate the usefulness of such a decision framework for 3D point cloud data. The proposed decision framework is generic and holds good for more than two interpolation techniques.