摘要
传统的Harris角点检测算法检测效果比较理想,也有较高的重复率,然而其角点提取的效果几乎完全依赖于非极大值抑制过程中阈值的设定,当参数不改变时,并不具有旋转不变性。文中提出了一种改进的Harris算法,通过二次非极大值抑制,有效地避免设置阈值这一难点;同时,在不改变任何参数的情况下,明显地减小了图像旋转前后角点提取的差异。通过对比实验表明,本算法简单易操作,提取角点非常有效,相较于传统的Harris算法具有更好的角点检测性能。
The traditional Harris comer detection algorithm not only has a good work,but also has a higher repetition rate. However, the whole comer detection process depends almost entirely on the comer extraction threshold during non-maxima suppression, and it＇s hard to achieve the rotational invariance if the parameters do not change. It presents an improved Harris algorithm, through the secondary maxima suppression to avoid setting the threshold value;at the same time, reduce the comer extraction differences around image rotation when there is not any change to the parameters. The experiment shows that this algorithm is simple and easy to operate, the comer points extrac- ted are very effective, compared with the traditional Harris algorithm has better comer detection performance.
出处
《计算机技术与发展》
2013年第2期11-14,共4页
Computer Technology and Development
基金
国家"863"高技术发展计划项目(2012AA011804)
关键词
HARRIS算子
角点
非极大值抑制
旋转不变
阈值
Harris operator
comer
non-maxima suppression
the rotational invariance
threshold
