基于Harris-SIFT算法的缝料视觉定位系统

Harris-SIFT Positioning Algorithm for Two-Dimensional Space Cloth

摘要:

在缝制行业中，为了提高缝制工艺的自动化程度，通常采用机械手实现对缝料的装卸工作。对于流水线上姿态和位置随机的缝料，采用传统工业机器手示教和预先编程的方式已无法满足工艺要求。本文在对传统SIFT（Scale-invariant feature transform）算法分析的基础上，设计了一种基于Harris-SIFT算法的缝料视觉定位系统。系统首先对视觉定位平台进行相机标定，建立图像几何模型和消除畸变；其次，将采集到的图像进行预处理操作，采用Harris-SIFT算法计算获取姿态信息；最后，利用定位平台对不同形状的缝料花样进行实验验证。实验结果表明，系统能够快速提取缝料的特征点并进行高精准匹配，其定位精度达到0.2mm，角度偏差小于0.15°，且运算速度相比传统SIFT定位系统提高了4-5倍，能够满足缝料工艺操作要求。

Abstract:

In the sewing industry, in order to increase the degree of automation of the sewing process, manipulators are usually used to realize the loading and unloading of sewing materials. However, for sewing materials with random postures and positions on the assembly line, the traditional methods of teaching and pre-programming by industrial robots can no longer meet the technological requirements. Based on the analysis of the traditional SIFT (Scale-invariant feature transform) algorithm, this paper designs a sewing material visual positioning system based on the Harris-SIFT algorithm. Firstly, the system performs camera calibration on the vision positioning platform, establishes a geometric model of the image and eliminates distortion; secondly, it preprocesses the collected images and uses the Harris-SIFT algorithm to calculate the posture information; finally, it uses the positioning platform to experimentally verify sewing patterns of different shapes. Experimental results show that the system can quickly extract the characteristic points of the sewing material and achieve high precision matching. Its positioning accuracy reaches 0.2mm, the angle deviation is less than 0.15°, and the calculation speed is 4-5 times higher than that of the traditional SIFT positioning system, which can meet the technique requirements that the operating process of the sewing materials demand.