利用Zernike矩计算储粮害虫图像倾斜角度

摘要

自动检测粮粒外部害虫是粮虫领域的研究热点.由于储粮害虫图像的一些无量纲的特征丧失了旋转不变性,为了利用这些无量纲的特征形成适于分类的最优知识库,必须计算储粮害虫图像的倾斜角度.分析了利用Zernike矩计算图像倾斜角度的原理,借助Zernike矩旋转不变性,提出了利用Zernike矩计算无明显直线的储粮害虫图像的倾斜角度的方法.试验证明,对于无明显直线的储粮害虫图像,利用Zernike矩能够很好地进行倾斜检测,倾斜角度计算准确度高,且该方法还具有对噪声不敏感的特性,对于其他无明显直线的图像也同样适用.%Automatic recognition for storage grain insects outside grain kernels has been a hotspot of research interests in the field of storage grain insects. Because some dimensionless characteristics in image of storage grain insects lost rotation invariance, the slanting degree of the storage grain insect's image must be computed in order to construct the optimal knowledge base suitable for classification. The theory of computing slanting degree of image based on Zernike moment has been analyzed. Because Zernike moment is rotationally invariant, the method based on Zernike moment was introduced, which computes the slanting degree of the stored-grain insect's image that does not have obvious linear characteristic. The experiment shows that the method can measure the slanting degree of the storage grain Insects image well with high accuracy. The method is insensitive to noise owing to Zernike moment's good anti-noise interference capability. The method is also applied in others image which does not have linear characteristic.