一种基于图元方向变形的人体模型生成新算法

摘要

提出了一种基于图元方向变形的人体模型生成算法.通过图元截线点以及图元方向变形的方式获得了不同部位的人体图元,从而很好地避免了模拟函数所产生的误差.同时,利用图元网格生成与平滑连接相结合的方式生成了不同人体体型.该方法既可以使生成的人体图元被重复使用,不同类型图元的组合又可以使模型的生成更有针对性.实验结果表明,与传统方法相比,本方法生成的模型更加逼真,与模板模型差异更小.另外,由于图元模型的可重用性,也可进一步提高模型生成的效率.%We present a human model generation algorithm based on orientation deformation on primitives.By using the primitive curve spots and orientation deformation,we can obtain human primitives of different human parts.At the same time,it can also eliminate displacement when some simulated functions are used.Meanwhile,different human models are generated through primitive mesh generation and smooth connection.This method can not only improve the reusability of generated human primitives,but generate some specific models via the combination of different types of primitives.Experimental results show that compared with traditional methods,the models generated by the proposed method are more real and have less difference with the template model.Furthermore,the generation efficiency of human models can be further improved because of the reusability of primitive models.