Skip to Main Content
In order to explore effectiveness of image inpainting technique for encoding image of coded-structured-light 3D measurement system, image inpainting method, image inpainting-reconstruction experiments, quantization and visual evaluation scheme were investigated. First, exemplar-based image inpainting method and its adaptability to encoding image were analyzed. Second, inpainting experiment of linear edge image was done, and quantization evaluation scheme based on straightness, parallelism and their composite errors was designed. Third, inpainting experiment of curved edge image was done, and visual effectiveness of reconstructed surface was evaluated. Experimental results indicate that composite error of broad-stripe image (0.33pixel) is smaller than that of narrow-stripe image (0.80pixel). Reconstructed surface based on inpanted encoding image has real visual effectiveness, while there are gross errors in it. Its availability in low-accuracy civil field should be further analyzed.