3-D Imaging and Inverse Labeling of Biological Samples Based on Reflectance-Guided Adaptive Fringe Projection

Ziqiang Wei, Sudong Ding, Yan Hu, Shuang Mu, Kehui Wang, Kun Gui, Shijie Feng, Qian Chen, Chao Zuo

Abstract: Vision-based 3-D measurement techniques play a crucial role in capturing and restoring the realism of scenes. However, when measuring objects with significant reflectance variance, the captured image is prone to saturation, leading to errors and incomplete or malformed 3-D data. This article presents an effective and robust adaptive fringe projection method specifically designed for measuring high dynamic samples with highlighted surfaces. Our approach iteratively reduces the light intensity at regions of the projected patterns corresponding to the highlighted surface in several steps, ensuring the full-field unsaturated intensity in the captured images. Then, the reflectance map is used to optimize the phase-shifted patterns further to reconstruct 3-D data, which has smoother contours than methods employing multiple exposures with numerous fringe images. By mapping the drawn lines on the sample image into the projected patterns, inverse labeling is performed by projecting positioning lines to guide the cutting of pathological samples. We carried out experiments on accuracy validation and comparison with other methods, and the results demonstrate that our proposed method accurately reconstructs complete 3-D profiles of highlight surfaces on biological samples and can project precise labeling lines with adjustable thicknesses and shapes.