虚拟现实的图像复原真实性优化仿真研究

摘要

This article presents an optimization method for authenticity of image restoration in virtual reality based on non-local mean theory.Firstly,our research confirmed appropriate reference plane according to mutual matching relation between image chips of virtual reality and worked out corresponding matrix from each image chip to reference plane.Then,the research converted two pieces of image chips waiting for registration into identical coordinate plane integrated with bilinear interpolation method to joint.Finally,we obtained weights among each pixel of image chip and built the model for image restoration integrating sparse regularization with non-local regularization.Experimental results show that the method can reserve texture detail information of virtual reality image.The effect of restoration is excellent.%对虚拟现实中图像复原的真实性进行研究,可以有效提升虚拟现实场景的视觉真实感.对虚拟现实的图像复原真实性优化,需要计算出图像各像素之间的近似度,得到图像片各像素间的权重.传统方法通过构建两个虚拟现实图像规则化约束项,来完成虚拟现实图像的复原,但忽略了图像像素间的权重,导致复原效果不理想.提出基于非局部均值理论的虚拟现实图像复原真实性优化方法.依据虚拟现实图像片间的相互匹配关系确定合适的参考平面,计算出每块图像片到参考平面的相应矩阵,结合双线性内插理论将待配准的两块虚拟现实图像片转换至同一坐标平面内进行拼接,得到图像片各像素间的权重,并组建稀疏与非局部规则化相融合的虚拟现实图像复原模型.实验结果表明,所提方法能够有效保留虚拟现实图像的纹理细节信息,且复原效果较好.