Feature extraction is an important technology in computer vision, Harris and SIFT algorithm in feature extraction are compared, and simulation experiments show that SIFT algorithm is better when dealing with the scale zooming and brightness variations of the images. According to the three-dimensional imaging principle, this paper reconstructs the extracting feature points of the SIFT algorithm based on three-dimension, and target disparity maps and three-dimensional reconstruction of the scene graph are obtained, which further illustrates the effectiveness of extraction of SIFT algorithm.%特征提取是计算机视觉中的一项重要技术，本文针对特征提取中的Harris算法和SIFT算法进行了比较，通过仿真实验证明，在处理尺度缩放、亮度变化的图像时SIFT算法更加优越。根据三维成像原理，本文对SIFT算法提取的特征点进行三维重建，获得了目标场景的视差图和三维重建图，进一步说明了SIFT算法的提取是行之有效的。
展开▼
