Abstract:

Aiming at the problem of high precision modelling of multi contour 3D stereo model, it was necessary to carry out the projection detection and separation to improve the image quality, because of the speckle and projection of the visual switching and attenuation of the light intensity, we proposed a projection detection algorithm based on the similarity characteristics of texture segmentation of multi contour 3D stereo vision images. Firstly, according to the known multi contour 3D model, the proposed algorithm made traversal modelling analysis for all the modelling point. Secondly, the model for modelling point
after traversal were reconstructed, and planar control basis functions with integral recursion polynomial were constructed. Finally, an improved  3D stereo vision image projection detection iterative equation was obtained by using the curve mixed function initialization detection model of multi contour image. We used similarity feature texture segmentation method to improve the projection detection of visual image, which solved the problem of inaccurate image projection detection. The simulation results show that the algorithm can effectively realize the separation of image projection detection, image retention is better, and improve the image quality.

Key words:  3D image, vision, projection, detection