高速的9×9尺寸模板Zernike矩边缘算子

A high-speed Zernike moments edge operator based on 9×9 masks

根据Zernike矩基本原理,提出了基于9×9模板尺寸的Zernike矩边缘算子。首先推导Zernike矩模板计算过程,然后计算出2个9×9模板,并选取二值图像和经过5次平滑的模糊图像为对象对新算子进行了测试。结果表明:对于理想的二值图像,新算子具有较强的边缘细化能力,且边缘检测精度也较高;而对于模糊图像,虽然新算子的边缘检测效果也较好,但应注意边缘图像表征与亚像素坐标数据间存在不一致的现象。最后,采用先粗后精定位思想,使新算子的运算速度大幅提高,耗时仅为0.19s。

Based on the Zernike moments, a high-speed Zernike moments edge operator based on 9×9 masks is developed.Firstly, the calculation steps of Zernike moments masks are derived,and then two 9×9 masks of the new edge operator are given.Two different images are chosen for the new operator testing：one is the original artificial binary image, and the other is the fuzzy image after 5 times of smoothing.The results show that the new operator has strong edge refining ability, and the edge detecting accuracy is satisfied.Although the fuzzy image testing also shows that the new operator has good detection result, it should be noted that there is mismatching between the edge image and the recorded sub-pixel coordinate data.Finally, the two step detection stragety is used to improve the operation speed.The running time of the new operater is no more than 0.19 s.