基于角点匹配的强力输送带接头识别算法

摘要

For accurate positioning of conveyor belt internal wire rope damage, a joint recognition algorithm based on corner feature vectors is proposed, taking the joint as datum point to select the complete image of joint. According to normalized product correlation algorithm, the Haar wavelet response distribution is calculated using the Harris corner points as feature points, each joint is identified according to the feature vector matching of joint image. Experiments show that, illumination, robustness, recognition rate and efficiency of the algorithm is favorable, realizing effective joint identification and fault positioning.%为精确定位输送带内部钢丝绳损伤，提出一种基于角点特征向量匹配的接头识别算法。该算法以接头为基准点，根据归一化积相关算法筛选出完整的接头图像，利用Harris角点作为目标特征点计算该特征点的haar小波响应分布，形成接头图像的特征向量并依据特征向量匹配识别出每个接头。实验表明，该算法不仅具有良好的光照、平移鲁棒性和识别率，而且速度快、效率高，可以有效解决接头识别和故障定位问题。