Comparative study of B-spline and cardinal spline with genetic algorithm for invariant shape object recognition

This paper presents an analysis technique that can be used with genetic algorithm to select a model shape, that has the best match with invariant input images. First, the dominant points are selected from the boundary of the object using Gaussian filtering. Secondly the control points and data points are computed using the B-Spline and Cardinal Spline respectively. The model shapes of objects are created from these points into the training database. Then the genetic matching technique is applied to find the best-matched model between the model shapes and input model. The results have been compared between the model from B-Spline and Cardinal Spline. The recognition rate for the B-Spline implementations are 94.5 % for rotated objects, 92.1 % for rotated and scaling objects, and for the Cardinal Spline implementations are 95.3 % for rotated objects, 93.5 % for rotated and scaling objects respectively.