摘要
研究图像分割精度质量问题,由于图像中存在噪声和伪影,造成区域边界不连续,影响提取质量。传统的图像分割算法提取的图像质量差。为了解决上述问题,提出了一种利用几何旋转不变性特征和直线特征的图像分割算法,有效地提高了图像分割的准确率。算法首先分别对图像提取SIFT和LOFO特征,并在两种特征基础上进行匹配,获得匹配值,然后算法将两个匹配结果有机的结合起来,得到最终的匹配值。仿真结果表明,提出的新的几何不变性原理图像分割算法能快速的有效的分割图像,不仅可以得到了比较高的分割精度,获得有效的提取。还大大减少了计算量,并能够改善图片分割的效率和质量。
The accuracy of image segmentation quality was studied. Because of too much emphasis on image seg mentation accuracy, the traditional image segmentation algorithms are of increased complexity of. To solve this problem, we proposed an image segmentation algorithm based on rotation invariance and linear feature, which effectively improved the accuracy of image segmentation. The algorithm first extracted LOFO and SIFT features respectively for the image, and matched on the basis of two features to get a match value. Then the algorithm combined the two matches to get the final match. The simulation results show that the proposed new image segmentation based on geometric invariant theory can be fast and effective for image segmentation, It has a relatively high segmentation accuracy, and an effective feature extraction ability, sgreatly reduces the amount of computation, and improve the efficiency and quality of image segmentation.
出处
《计算机仿真》
CSCD
北大核心
2012年第1期257-259,388,共4页
Computer Simulation
关键词
几何不变性原理
图像分割
特征提取
特征匹配
Geometric invariance principle
Image segmentation
Feature extraction
Feature matching
