基于自适应基元的导管测量方法

Tubes are widely used in automobile, ship, aerospace and other fields. It is an important means to measure the spatial shape of the processed tube to ensure its manufacturing accuracy and assembly reliability. At present, the measurement methods commonly used in engineering rely too much on human operation; In recent years, the computer vision based measurement methods cannot take into account the efficient and accurate measurement of the straight and arc segments of the tube. The proposed tube measurement method based on adaptive primitives aims to solve the difficult problem of tube measurement. According to the shape characteristics of the tube, the cylinder is used as the geometric element to initialize and reconstruct the tube by matching the contour of the tube in the image, and the straight and arc segments are identified. The arc segments with different bending radii are measured with different lengths of the primitive to promise the corresponding measurement accuracy. Based on the training data, the support vector machine (SVM) is used to establish the primitive parameter prediction model. According to the different bending radii of the arc segments, on the premise of satisfying the measurement accuracy, the longest primitives can be predicted to ensure the efficient and accurate measurement of the straight and arc segments of the tube.