基于Hough投票空间的椭圆图像特征亚像素提取方法

Sub-pixel ellipse feature extraction method based on Hough voting space

椭圆检测是机器视觉和模式识别的基本任务。首先在椭圆中心初定位阶段,通过参数空间的少量投票,依据均值漂移技术获取粗略的椭圆中心,然后再由精提策略实现椭圆中心的精确像素级定位,并采用插值方法进一步获取椭圆中心亚像素坐标。仿真及真实图像的实验结果表明,所提出的方法同时具备了椭圆检测的鲁棒性和中心定位的精确性。

Ellipse detection is the basic task of machine vision and pattern recognition. The proposed algorithm in the early stages via the mean shift technique obtains a rough center of the ellipse through parameter space of a small amount of voting for the center of the ellipse. Then the accurate pixel-level center of the ellipse is obtained by refining strategies, and further in- terpolation method is used to get sub-pixel ellipse center coordinates. Experimental results for simulated and true image show that the proposed method not only achieves robust ellipse detection, but also receives the accurate center location.