【摘要】 针对在焊接过程中工件易变形、光照干扰等情况下不能精确定位焊缝的问题,提出一种基于方向梯度特征的随机蕨丛算法。首先建立焊缝模型分类器,并用其在测试样本中得到焊缝部分。其次,采用随机蕨丛算法得到所有训练样本的特征类型并计算特征点的平均坐标差值。最后,判断测试样本所属类型,根据坐标偏移量求出特征点具体坐标值来实现定位。实验结果表明:所提出的算法在保证较高精确定位的同时,还具有很好的时效性。更多还原

【Abstract】 Aiming at the problem that is not easy to accurately locate the weld seam when the workpiece is deformed or the situation of light interference,an algorithm based on histogram of oriented gridients is proposed for Random ferns. Firstly,the weld model classifier is established and the classifier is used to get the weld part in test samples.Secondly,the Random ferns is used to get the feature types of all the training samples,the feature points’ average coordinate difference is calculated. Finally,we can determine the type of test samples,get the specific coordinate value to achieve the positioning according to the coordinate offset. The experimental results show that our algorithm can ensure the high precise positioning and good timeliness.更多还原