基于SURF特征点的金属罐图案检测算法

Detective Algorithm of Metal Can Patterns Based on SURF Feature Points

为快速、准确地检测出金属罐生产过程中出现的图案倒置问题,提出一种基于SURF特征点的金属罐图案检测算法.该算法通过比较金属罐图像与模板图像间匹配特征点的位置关系,实现金属罐图案方向的判别.首先利用SURF算法分别提取分区域处理后的待检测图像和模板图像的特征点;再利用双向KNN算法和RANSAC算法进行特征点匹配;最后计算匹配特征点的位置关系,并判别金属罐方向.实验表明,该算法能够有效地实现金属罐图案倒置的检测,可以达到每分钟800罐的检测速度.

In order to detect metal can pattern inversion problem quickly and accurately,a detective algorithm of metal can patterns based on SURF feature points was proposed. The algorithm can recognize the metal can pattern＇s direction by comparing the positional relationship of the matched feature points between the detected metal image and the template image.Firstly,SURF algorithm is used to extract the feature points of the detected image and the template image which are divided into different regions.Secondly,the feature points are matched using bidirectional KNN algorithm and RANSAC algorithm.Finally,the metal can pattern＇s direction is given by calculating the matched feature points＇ positional relationship.The experiment results indicate that the algorithm can solve the metal can pattern inversion problem and the detection speed is 800 cans per minute.