高斯插值哑像素边缘检测算法的优化

针对传统的高斯插值算法运用在图像检测上定位不准确，定位精度随着所选择定位角度的不同而不同，效率低等问题，对高斯插值亚像素边缘检测算法进行了优化．首先，通过LOG算子获得像素级的边缘后，进行Hough变换，得到图像边缘的斜率和在垂线方向相应点，并设置新的坐标系；然后，利用泰勒级数插值法在新的坐标系下获取灰度值的方向梯度；最后，对变换后的亚像素梯度方向进行高斯插值计算．实验结果表明：优化后的方法比传统的算法可以得到更好的精度，更高的效率，且具有方向不变性，验证算法的精度达到0．05个像素，有效地提高了图像边缘提取的精确度．且将该算法运用于刀具几何参数的检测和视觉设备效果很好．

The Optimization of Algorithm with Sub-pixel Edge Detection Gauss Interpolation

The traditional Gaussian interpolation algorithm used in image detection is not so accurate. The positioning accuracy changes with the different angles of positioning location. It also has the problem of low efficiency. Because of those above problems, the algorithm of Gaussian interpolation sub -pixel edge detection is optimized. This paper presents an improved algorithm： After obtaining pixel- accuracy edge of object to be measured by LOG operator, Hough transform is used to get the curve slope of the boundary line and the normal of the corresponding point on the edge; Taylor series interpolation in the normal direction is performed to obtain the gray values in the direction of the gradient under new coordinate system; Finally, sub-pixel relocation in the gradient direction is performed based on the Gaussian interpolation algorithm. Experimental results show that the improved method can get much better precision, higher efficiency than the traditional algorithm. The algorithm has an advantage of direction invariance. The precision of 0. 05 pixels effectively improves the accuracy of image edge detection. With very good effect, The method is applied to the detection of tool geometry and visual equipment.