基于SIFT算法的图像特征点提取和匹配研究

Research of features extraction and matching in the image based on the SIFT algorithm

目的:把SIFT算法应用在牙齿模型图像上,检测牙齿图像的特征点。方法:首先采用高斯差分算子DoG搜索整个图像的尺度和位置信息,从而确定具有代表性尺度、方向的特征点。基于其稳定性选择关键点,得到一个详细的模型以确定每个候选点的合适位置和范围。基于局部图像梯度方向信息将方向矢量和关键点对应起来。在选定范围内的每个关键点周边区域测量局部图像梯度,并采用KNN算法进行特征匹配。结果:通过大量的实验和与其他特征提取方法相比较,该方法能有效地检测牙齿模型图像的特征,并为牙齿模型三维重建提供有效的参数。结论:在口腔医学牙齿模型图像中应用SIFT算法,有助于牙齿模型的三维重建。

Objective：To detect the features of the teeth image effectively,SIFT algorithm was preformed to use in the image of teeth model. Methods：This approach identified potential interest points by searching over all scales and image locations with the difference-of-Gaussian function. Keypoints were selected based on their stability and a detailed model was obtained to determine location and scale of the candidate location. Assign one or more orientations to each keypoint location based on local image gradient directions. The gradients of local image in region around each keypoint were measured,and the features were matched by KNN algorithm. Results：This method can detect the features effectively and offer some available parameters for 3D reconstruction of the teeth model through lots of experiments and comparing with other feature extraction algorithms. Conclusion： SIFT algorithm performed in the image of teeth model may contribute to 3D reconstruction of the teeth.