结构光技术是一种简单快速的非接触式三维测量技术，被广泛应用于工业检测、生物医学、人体测量、自动驾驶和数字娱乐等多个领域。二值离焦可以突破结构光系统中光学图案投影的速度瓶颈，是高速、实时结构光系统中最受欢迎的方法之一。本文以手术辅助引导和工业轮廓检测为场景，设计实现了基于二值离焦的高速彩色投影三维重建系统，同时针对离焦场景下的投影仪畸变修正提出了一种后处理方法，并提出适用二值条纹的预畸变处理方法。彩色投影技术是一种提高相机帧利用率的技术，可以有效提升系统帧率。为了提高三维重建系统的速率，并应对物体表面颜色不理想的情况，本文结合二值离焦和彩色投影技术设计了一种投影编码方案，将三个频率的条纹分布在三个不同颜色通道中，搭建了一套离焦彩色三维重建系统，并为系统配置了软件平台。系统在缺失单个颜色信道信息时也能完成重建，从而减少非理想颜色场景的影响，并且相对于现有离焦彩色投影方法能减少1帧白色投影，在点云帧数超过每秒120帧的同时精度达到0.85 mm级，通过提高系统帧数减少运动带来的误差。投影仪的非线性误差是结构光系统的主要误差之一，为了提高精度、简化计算复杂度，本文提出了一种适用于二值离焦系统的投影仪畸变后处理方法。投影仪非线性模型下的相位曲线在相机平面中可以被低阶多项式良好近似，基于极线校正的原理，通过近似多项式去修正非线性误差，拥有比迭代修正更低的计算复杂度的同时能够保留求解精度，并能够适用于离焦场景下存在相位失真的条纹，对相位进行一定修正，在提高系统处理速度的同时可以保证系统精度，实验表明提出方法相比于现有的快速处理方法有着20 %以上的误差减少。传统的预畸变方法在离焦条纹二值化的条件下会严重破坏条纹结构，本文提出了一种适用于二值化场景的预畸变方法，可以减少条纹结构的失真，将畸变信息提前嵌入投影图案中，无需在解码侧额外增加步骤，减少了解码计算复杂度，降低了硬件速度的要求。
Structured light technology, as a simple and rapid non-contact 3D measurement method, has been widely applied in industrial inspection, biomedical fields, anthropometry, autonomous driving, and digital entertainment. Binary defocus technology can overcome the speed limitation of optical pattern projection in structured light systems, making it one of the most popular approaches for high-speed real-time structured light systems. Focusing on surgical assistance guidance and industrial contour detection scenarios, this paper designs and implements a high-speed color projection 3D reconstruction system based on binary defocus. We propose a post-processing method for projector distortion correction in defocused scenarios and develop a pre-distortion processing method suitable for binary fringe patterns.Color projection technology enhances camera frame utilization efficiency, effectively improving system frame rates. To increase the speed of the 3D reconstruction system and address challenges posed by non-ideal object surface colors, this paper combines binary defocus with color projection technology to design a projection coding scheme. By distributing three-frequency fringes across different color channels, we established a defocused color 3D reconstruction system with supporting software platform. The system maintains reconstruction capability even with missing single-color channel information, thereby reducing impacts from non-ideal color scenarios. Compared with existing defocused color projection methods, our approach eliminates one frame of white projection while achieving point cloud frame rates exceeding 120 fps with 0.85 mm-level accuracy, effectively mitigating motion-induced errors through enhanced frame rates.To improve precision and reduce computational complexity, we propose a post-processing method for projector distortion correction specifically adapted to binary defocus systems. The phase curve under projector nonlinear model can be well approximated by low-order polynomials in the camera plane. Based on epipolar rectification principles, our method uses approximate polynomials to correct nonlinear errors, achieving comparable accuracy to iterative correction methods with significantly lower computational complexity. This approach remains effective for phase-distorted fringes in defocused scenarios, enabling both processing speed improvement and accuracy maintenance. Experimental results demonstrate over 20% error reduction compared to existing fast processing methods.Traditional pre-distortion methods severely disrupt fringe structures under binary defocus conditions. We propose a binary pattern-adapted pre-distortion method that minimizes structural distortion while embedding distortion information into projected patterns. This innovation eliminates additional decoding steps, reduces computational complexity in decoding, and lowers hardware speed requirements.
