Search

Image matching is the research basis of many computer vision problems, such as intelligent driving, object recognition and structure from motion. However, the traditional feature-based image-matching results are usually very sparse and unevenly distributed for wide baseline or weakly textured images. Implementing an efficient and robust image-matching technology is a challenging task. To solve these problems, we propose an efficient extractor and binary descriptor based on superpixels and a modified binary robust independent elementary features (BRIEF) descriptor called FSRB. FSRB can improve the computational efficiency, number of matches, feature distribution and robustness of feature-based image matching. In theory, FSRB is rotation-, scale-, affine-, distorted-, and intensity-invariant. A comprehensive performance evaluation of FSRB is performed. The experimental results show that our method can effectively obtain many matches for different types of images. Compared with state-of-the-art algorithms, our method performed very well in terms of the number of correct matches (which increased by 2-5 times), time consumption, matching accuracy, matching success rate and feature repetition rate. In addition, our method is applied to sparse 3D reconstruction of multiview images, and satisfactory results are obtained.

In the field of geoinformation science, multiview, image-based 3D city modeling has developed rapidly, and image depth estimation is an important step in it. To address the problems of the poor adaptability of training models of existing neural network methods and the long reconstruction time of traditional geometric methods, we propose a general depth estimation method for fast 3D city modeling that combines prior knowledge and information propagation. First, the original image is downsampled and input into the neural network to predict the initial depth value. Then, depth plane fitting and joint optimization are combined with the superpixel information and the superpixel optimized depth value is upsampled to the original resolution. Finally, the depth information propagation is checked pixel-by-pixel to obtain the final depth estimate. Experiments were conducted using multiple image datasets taken from actual indoor and outdoor scenes. Our method was compared and analyzed with a variety of existing widely used methods. The experimental results show that our method maintains high reconstruction accuracy and a fast reconstruction speed, and it achieves better performance. This paper offers a framework to integrate neural networks and traditional geometric methods, which provide a new approach for obtaining geographic information and fast 3D city modeling.