利用光强信息的结构光图像轮廓提取修正方法

Profile extraction of structural light image and its correction based on light intensity

在用激光三角法对同质物体进行三维探测中,可能因为条纹缺陷等原因出现轮廓提取误差。这种误差用诸如中心线或者权重法等一般轮廓提取方法无法得到正确的结果,因此提出一种利用光强信息进行修正的新方法。通过自行搭建的试验装置获得原始光强图像,由中心线法提取带有误差的轮廓,再使用光强分布对其进行修正。根据物体表面光强与表面法线方向的关系,用逐次逼近的方法验证物体表面的法线方向。通过理论推导计算以及和之前获得的轮廓进行比较,得到修正值。在测量标准高度为 40.1mm 的物体表面时对出现的条纹缺失进行修正,与传统方法相比,测量结果从 38.4mm 改善到 39.4mm,误差从 4.23%减小到 1.75%。

During 3-D detection of objects with same reflectance by laser triangulation, profile extraction error may appear because of loss of stripes. Due to this error, correct results cannot be obtained by use of common-used profile extraction methods such as center line method or weighting method. So a new method correcting with light intensity information is proposed. The profile with errors are extracted by center line method and then corrected by use of light intensity distribution. The normal direction on the object surface can be verified by successive approximation based on the relationship between light intensity and normal direction on object surface. The updating values can be obtained through theoretical derivation, calculation and comparison with the already obtained profile. Experimental measurement and correction for laser stripe loss appeared on the surface of the measured object with practical height of 40.1 mm are carried out. Compared with traditional methods, its measuring result became from 38.4 mm to 39.4 mm, which approaches the true value, and its error has been reduced from 4.23% to 1.75%.