首页|期刊导航|液晶与显示|面向曲面屏的透视变换无畸变3D渲染与显示方法

面向曲面屏的透视变换无畸变3D渲染与显示方法OA

Distortion-free 3D rendering and display method using perspective transformation for curved screens

中文摘要英文摘要

针对自发光曲面屏因物理弯曲导致的图像几何畸变问题,本文提出一种基于透视变换的自适应分段曲面屏无畸变3D渲染与显示方法,使用户观看曲面屏上的渲染场景等效于无屏直接观察三维场景.将连续曲面屏近似为多段平面屏的拼接结构,基于透视变换原理与已知视点位置,反转物像关系并利用透视变换矩阵计算各子屏显示图像;采用自适应区域划分策略,通过Adam算法最小化曲线与拟合折线间的二维均方误差,确定最优划分;在拼接边界处设置重叠区Δw并采用线性插值权重函数进行平滑处理,消除接缝失真.Blender 4.0仿真表明,本文方法处理后图像的峰值信噪比(PSNR)达到13.23 dB,相比直接渲染的7.65 dB提升了5.58 dB,超过12 dB的畸变可接受阈值;采用30段(m=30)平面屏拟合曲面,兼顾高PSNR与计算量.本方法可显著降低曲面屏显示畸变,适用于大型户外曲面LED及算力受限的柔性消费电子设备,为裸眼3D技术在曲面显示载体上的应用提供了可行技术路径.

To eliminate geometric distortion on self-luminous curved screens caused by physical curvature,this paper proposes an adaptive segmented distortion-free 3D rendering and display method for curved screens based on perspective transformation,enabling viewers to see the rendered scene identically to directly viewing the 3D scene without the screen.The curved screen is approximated as multiple planar segments.Using perspective transformation and the known viewpoint,the object-image relationship is reversed,and the perspective matrix computes images on each segment.Adaptive partitioning minimizes the 2D mean square error between the curve and fitted polyline via the Adam algorithm.Stitching boundaries are smoothed with overlapping regions Δw and linear interpolation.Blender 4.0 simulations show the proposed method achieves a PSNR of 13.23 dB,which is 5.58 dB higher than direct rendering(7.65 dB),exceeding the 12 dB distortion threshold.Using m=30 planar segments balances high PSNR and computational cost.This method significantly reduces distortion on curved screens and applies to large outdoor curved LEDs and flexible consumer electronics,offering a feasible path for glasses-free 3D on curved displays.

丁健;张书赫;曹良才

清华大学 精密仪器系,北京 100084清华大学 精密仪器系,北京 100084清华大学 精密仪器系,北京 100084

信息技术与安全科学

3D渲染曲面屏无畸变显示透视变换区域划分

3D renderingcurved screendistortion-free displayperspective transformationregion partition

《液晶与显示》 2026 (7)

970-980,11

国家自然科学基金(No.62441613)Supported by National Natural Science Foundation of China(No.62441613)

10.37188/CJLCD.2026-0082

评论