基于图像识别技术的机械手臂过角度判断

Based on Image Recognition Technology of the Mechanical Arm Angle Judgment

针对传统方法容易受到噪声及外界条件的干扰,判断时误差较大、无法有效实现机械手臂过角度的判断的问题。提出采用机械手臂过角度判断方法,通过最大类间方差法计算机械手臂图像的自适应阈值,完成图像分割。利用中值滤波法对图像进行滤波,完成对图像的膨胀处理。将机械手臂图像边缘点梯度向量作为匹配信息对相似度量进行计算。依据曲面拟合原理对边缘梯度方向进行计算,对机械手臂图像进行求导,对图像的梯度进行计算,判断图像的阈值与极大值抑制获取像素级的边缘位置,将像素级边缘点某范围内的梯度值作为拟合信息,完成梯度曲面拟合,通过最近邻法实现机械手臂过角度的判断。仿真结果表明,改进方法进行机械手臂过角度判断时的平均实测角度更接近实际值;平均误差和所需时间都有了较大的改善,具有较好的实用性。

The traditional method is easily affected by the noise and interference of external conditions,so that the judgment error is big,and it is unable to effectively realize the over- angle judgment of mechanical arm. An over-angle judgment method of mechanical arm is proposed. By using maximum between- cluster variance method,the adaptive threshold of the mechanical arm image is calculated,and the image segmentation is accomplished. The image is filtered by median filtering to complete the expansion processing of image. The edge point gradient vector of mechanical arm image is as matching information to calculate similarity measurement. Based on the principle of surface fitting,the edge gradient direction is calculated. The mechanical arm image is made derivation,and the image gradient is calculated,the image threshold and maximum suppression are determined,and the edge position of the pixel level is obtained. The gradient values of pixel level edge points in a range are as fitting information,and the gradient surface fitting is completed. Through the nearest neighbor method,the over- angle judgment of mechanical arm is achieved. The simulation results show that the average measured angle using the improved method used for the over angle judgment of mechanical arm is more closed to the actual value,the average error and the required time have been greatly improved,and it has good practicability.