摘要
矩技术是描述图像特征的一种有效方法,在图像处理的各个领域得到了广泛应用。Zernike矩是一种连续正交矩,可以描述图像形状的全局特征,其最重要的性质是具有旋转不变性。Zernike矩能够构造任意的高阶矩,较容易地从矩特征中对图像进行重构,具有最小的信息冗余,并且对噪声不敏感、鲁棒性好。目前,在医学图像处理中Zernike矩较多应用于边缘检测和图像配准领域。Zernike矩可以达到亚像素级的边缘定位,且抗干扰性强;基于Zernike矩的配准代价函数降低了需要对比的自由度对数,从而显著减少了迭代次数;利用Zernike矩作为图像特征点描述子的算法,在图像存在比例缩放、旋转等情况下具有良好的配准效果。
The moment technique is an effective method for image feature description and is widely used in various fields of image processing. Zernike moments are one kind of continuous orthogonal moments. They can be used to describe the global shape features of the image. The most significant property of Zernike moments is rotation-invariance. Any high order Zernike moments can be constructed. And image reconstruction from Zernike moments can be easily achieved. Zernike moments have the minimum information redundancy and are robust and insensitive to noise. In medical image processing, Zernike moments are mainly used in edge detection and image registration. Literature survey shows that sub-pixel level edge position can be achieved based on Zernike moments and the algorithms have high noise immunity. The registration merit functions based on Zernike moments reduce the degrees of freedom needed to be optimized, and the number of iteration necessary for registration can be significantly reduced. Algorithms with Zernike moments as feature point descriptor show good registration results in the presence of image scale and rotation.
出处
《中国医学装备》
2012年第9期61-64,共4页
China Medical Equipment