Research on Space Positioning Technology of Microlens Array Based on Multi-Camera

Abstract

The light field camera is a novel optical device capable of capturing complete light field information in space, and it is widely applied in fields such as computer vision and computational imaging. The microlens array is an important component of the light field camera, and its alignment accuracy with charge coupled device is a key factor affecting the imaging performance of the light field camera. Due to the transparency of microlens array’s material and its complex surface structure, it is challenging to achieve high-precision spatial positioning of it through visual methods, which is a key technological bottleneck for realizing the automation of light field camera alignment and adjustment. To address this issue, this paper proposes a multi-camera-based microlens array spatial positioning technology. This technology enhances the multi-camera calibration method, presents an full contour recognition and feature point extraction algorithm, and proposes a two-step optimized spatial positioning algorithm based on multi-view stereo vision to achieve high-precision spatial positioning of microlens array, enabling observation of small displacements and rotations of it. Experimental validation shows that the system established in this paper has an average positioning error of less than $5~\mu $ m for translation motion and less than 30” for rotation motion, meeting the accuracy requirements for light field camera alignment.

Keywords

• Edge extraction
• MLA spatial positioning
• multi-camera calibration
• positioning accuracy verification