原始Zernike矩边缘算子抗干扰性能

针对Zernike矩算子在边缘模糊、随机噪声存在情况下的检测性能进行了研究，以获得该算子的抗干扰性能.通过分析Zernike矩算子的各阶矩计算过程，得知将像素灰度与模板进行卷积便可获得Zernike矩，这样的计算有利于减小随机干扰的影响.利用人工合成的二值图像进行了边缘检测性能测试，结果表明，对于模糊噪声图像，由于边缘的过渡特征，Zernike矩算子检测具有一定精确性，但会在某种程度上存在边缘细化能力、定位精度下降的问题；而对于随机噪声图像，当最大值在40以内时，Zernike矩算子具有较强的抗噪能力；若随机噪声超出此范围，则检测效果急剧恶化.

Noiseproof feature of original Zernike moment edge operator

In order to obtain the noiseproof feature of Zernike moment operator, the detection performance of the operator was investigated with existing indistinct boundary and random noise. Through analyzing the calculation process of Zernike moment operator, the Zernike moment could be obtained by the convolution computation of pixel gray with template. And thus, the effect of random noise could be reduced. In addition, the edge detecting performance of the artificial binary image was tested. The results show that for the fuzzy noise image, the detection of Zernike moment operator possesses a certain accuracy due to the transitional feature of the edge. However, the decline in both edge thinning ability and positioning accuracy appears. For the random noise image, the Zernike moment operator exhibits the stronger noiseproof ability when the maximum value of random noise is below 40. The detection effect will rapidly deteriorate when the maximum value of random noise is beyond this range.