基于机器视觉的汽车仪表指针检测方法

Detection Method of Automobile Instrument Pointer Based on Machine Vision

针对目前汽车指针式仪表读数识别由人工检测,导致错检率高的问题,提出一种基于机器视觉的指针检测方法。首先采用最小灰度相减法从两张背景相同的图像中提取无指针图像,以此为模板,让待检仪表与之进行匹配,矫正位姿。然后通过剪影法提取出仪表指针,二值化,闭运算后将指针用最小二乘法拟化成椭圆,利用椭圆长轴与水平线的夹角计算偏转角度。最后与标准表的偏转角度做比较,从而判断是否合格。实验结果表明,该方法可有效解决对于工艺偏差或设备轻微抖动造成的指针偏转测量有误的问题。算法稳定可靠,误差较小。

In view of the current manual detection of pointer readings in automobiles,which leads to high false detection rate,this paper proposes a pointer detection method based on machine vision.First,the minimum grayscale subtraction method is used to extract the no-pointer image from two images with the same background,and this is used as a template to allow the to-be-checked instrument to match with it to correct the pose.Then,the pointer of the instrument is extracted by the silhouette method,and the pointer is binarized.After the closing operation,the pointer is converted into an ellipse by a least squares method,and the deflection angle is calculated by using the angle between the major axis of the ellipse and the horizontal line.Finally the method is compared with the deflection angle of the standard table to judge whether it is qualified.The experimental results show that this method can effectively solve the problem of wrong hand deflection measurement caused by process deviation or slight device jitter.The algorithm is stable and reliable with less error.