Learning Two-View Correspondences and Geometry via Local Neighborhood Correlation

Seeking quality feature correspondences (also known as matches) is a foundational step in computer vision. In our work, a novel and effective network with a stable local constraint, named the Local Neighborhood Correlation Network (LNCNet), is proposed to capture abundant contextual information of each correspondence in the local region, followed by calculating the essential matrix and camera pose estimation. Firstly, the k-Nearest Neighbor (KNN) algorithm is used to divide the local neighborhood roughly. Then, we calculate the local neighborhood correlation matrix (LNC) between the selected correspondence and other correspondences in the local region, which is used to filter outliers to obtain more accurate local neighborhood information. We cluster the filtered information into feature vectors containing richer neighborhood contextual information so that they can be used to more accurately determine the probability of correspondences as inliers. Extensive experiments have demonstrated that our proposed LNCNet performs better than some state-of-the-art networks to accomplish outlier rejection and camera pose estimation tasks in complex outdoor and indoor scenes.

Relative Camera Pose Estimation using Synthetic Data with Domain Adaptation via Cycle-Consistent Adversarial Networks

Learning-based visual localization has become prospective over the past decades. Since ground truth pose labels are difficult to obtain, recent methods try to learn pose estimation networks using pixel-perfect synthetic data. However, this also introduces the problem of domain bias. In this paper, we first build a Tuebingen Buildings dataset of RGB images collected by a drone in urban scenes and create a 3D model for each scene. A large number of synthetic images are generated based on these 3D models. We take advantage of image style transfer and cycle-consistent adversarial training to predict the relative camera poses of image pairs based on training over synthetic environment data. We propose a relative camera pose estimation approach to solve the continuous localization problem for autonomous navigation of unmanned systems. Unlike those existing learning-based camera pose estimation methods that train and test in a single scene, our approach successfully estimates the relative camera poses of multiple city locations with a single trained model. We use the Tuebingen Buildings and the Cambridge Landmarks datasets to evaluate the performance of our approach in a single scene and across-scenes. For each dataset, we compare the performance between real images and synthetic images trained models. We also test our model in the indoor dataset 7Scenes to demonstrate its generalization ability.

LARGER RECEPTIVE FIELD BASED RGB VISUAL RELOCALIZATION METHOD USING CONVOLUTIONAL NETWORK

Visual Relocalization is a key technology in many computer vision applications. Traditional visual relocalization is mainly achieved through geometric methods, while PoseNet introduces convolutional neural network in visual relocalization for the first time to realize real-time camera pose estimation based on a single image. Aiming at the problem of accuracy and robustness of the current PoseNet algorithm in complex environment, this paper proposes and implements a new high-precision robust camera pose calculation method (LRF-PoseNet). This method directly adjusts the size of the input image without cropping, so as to increase the receptive field of the training image. Then, the image and the corresponding pose tags are input into the improved LSTM-based PoseNet network for training, and the Adam optimizer is used to optimize the network. Finally, the trained network is used to estimate the camera pose. Experimental results on open RGB dataset show that the proposed method in this paper can obtain more accurate camera pose compared with the existing CNN-based methods.