This article introduces a drone image stitching based on mesh-guided deformation and ground constraint, which can closely match the characteristics of images and achieve precise registration and acquire ideal stitching effect. The traditional methods use the homography model to align the image, which causes artifacts in the result of stitching the images with parallax. To overcome this situation, the image is divided into meshes and the mesh vertices of the target image are used to guide the warping. A new energy function is designed to represent the deformation characteristics of the image. We propose a new alignment term by using local homography and a local scale term by using the edge information of the mesh. The established mesh-guided deformation model can overcome image parallax caused by some external factors and eliminate the ghostly parts of the result. Moreover, imaged scene is not effectively planar and some fluctuations exist in the scene of the images, which will distort the stitching result. We propose a ground constraint with the ground plane as the main plane to reduce projection distortions in non-overlapping areas between images. Finally, the method of creating groundtruth is proposed, which can evaluate the naturalness of results and make comparison more reasonable. Several sets of challenging drone images are tested, and the experimental results show that our stitching system has good results.

This letter presents an air quality forecasting method, whose main strength is that the prediction accuracy and reliability can be improved effectively based on observed data of multiple sensor nodes. In our solution, for a certain sensor node, several spatial correlated neighboring sensor nodes are first selected according to the mutual information among nodes. Then time series data of these nodes are concatenated and fed into a gated recurrent unit (GRU) network to train the model for air quality forecasting. The proposed method is evaluated on the Intel Lab dataset and achieves better performance with about 14% and 5% reductions in terms of mean absolute error (MAE) and root mean square error (RMSE) compared to single node-based forecasting. Besides, the feasibility of the proposed method has been validated through a practical application of indoor air quality monitoring.

Image stitching stitches multiple overlapping images into a seamless image according to the corresponding geometric relationship between the reference and source images. In this article, a parallax-tolerant image stitching method based on robust elastic warping is integrated in the stitching of drone images, and locality preserving feature matching is used to effectively remove outliers from a set of putative feature correspondences. Our method can avoid the ghost issue in traditional methods when the scene suffers large view point changes. The method can be divided into three stages, namely, locality preserving feature matching, robust elastic warping, and global projectivity preservation. First, a set of high-precision matched feature points are provided for a pair of drone images, where the locality preserving matching is used. Second, the robust elastic warping function eliminates the parallax error, and the input image is distorted according to the calculated deformation on the nongrid plane. Finally, the global projectivity preserving method is applied to obtain high-precision resultant panoramas. Experiments on several sets of drone images demonstrate that our method can generate better panoramas over the state-of-the-art competitors.

This letter proposes a lightweight distributed time series anomaly detection method for environment monitoring sensor networks. To overcome environmental disturbance and ensure computation efficiency, for each sensor node, the observed data are converted into symbol sequences, and then trained through a self-learning model to obtain the local detection threshold. Within the framework of federated learning (FL), the global threshold is calculated using an energy-weighted method based on the local thresholds, and then used to continuously update the detection thresholds of sensor nodes. After training, any of sensor nodes independently determines whether an anomaly occurs by its own detection threshold. The proposed method has been successfully applied to indoor environment monitoring and experimental results demonstrate that our method has obvious advantages in terms of detection accuracy and energy efficiency.