由于变电站巡检机器人停靠位置和云台转动的偏差，拍摄所获得的图像中仪表位置会有较大不同。为解决图像中仪表的定位问题，文中提出了一种快速地利用边界提取椭圆形目标的方法，可有效地解决图像中圆形目标的提取。原有最小二乘椭圆拟合算法对图像边界上所有样本点都参与运算，所以会对椭圆拟合的最后结果产生偏差且耗时较长。针对这种情况，采取边界的最小二乘拟合算法，依次取图像边缘提取后的边界，逐段拟合椭圆，并对拟合出的椭圆进行评估，选取适合待检测目标的椭圆区域，具有快速高效、定位准确等特点。最后，通过合成图像和实际图像的应用验证了算法能够拟合出具有高精度的椭圆，能够有效地处理仪表表盘的提取。%It may have a certain bias between the device in the template image and new acquired image because of the deviation of the motion and mechanical error caused by PTZ(Pan-Tilt-Zoom) for mobile inspection robot. So how to locate the meter in a new acquired image is need to be solved. Present a fast elliptical target extraction method based on boundary that can effectively solve the extraction of the circular target in the image. The least squares ellipse fitting algorithm,all the sample points are involved in operations,so the final out-come produces the ellipse fitting bias and takes a long time. Adopt the least squares fitting algorithm based on the boundary,take the boundaries of the image edge extraction,piecewise fitting ellipse,ellipse and fitting out of the assessment,select the appropriate target to be detected elliptical area. Verifying the algorithm in the synthetic and actual image applications can fit oval with high precision,which can effectively deal with the extraction of the instrument dial.
展开▼
