研究目的
To propose an effective orthoscopic three-dimensional reconstruction via adjustable depth position in integral imaging system, overcoming the pseudoscopic problem and image degradation of conventional methods.
研究成果
The proposed method effectively reconstructs orthoscopic 3D images with faster reconstruction times and higher image quality compared to conventional methods, as demonstrated through experiments showing reduced pseudoscopic issues and improved PSNR, making it suitable for real-time applications like depth estimation and object recognition.
研究不足
The proposed method assumes the index s for reconstruction planes is a positive integer, leading to ray mismatch errors for non-integer values; it may be limited by the resolution and setup of the virtual pinhole array.
1：Experimental Design and Method Selection:
The methodology involves a computational orthoscopic 3D reconstruction using a one-step ray tracing method with adjustable depth positions, based on flexible ray tracing algorithms derived from smart pixel mapping techniques.
2：Sample Selection and Data Sources:
A 3D scene composed of two fishes and a coral at depths of 12 cm, 36 cm, and 72 cm is used, with a captured original elemental image array (EIA) of 1500x1500 pixels from a virtual pinhole array.
3：List of Experimental Equipment and Materials:
A virtual pinhole array with pitch of
4：02 mm, CCD for pickup, and computational setup for ray tracing. Experimental Procedures and Operational Workflow:
The pickup stage uses a micro-lens array and CCD to capture the EIA; reconstruction involves direct pixel mapping from the EIA to reconstruction planes using ray tracing equations.
5：Data Analysis Methods:
Performance is evaluated using reconstruction time and peak signal-to-noise ratio (PSNR) calculations, with depth estimation via 3D nonlinear correlation.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
