Mingyao Ai

A Robust Photogrammetric Processing Method of Low-Altitude UAV Images

Low-altitude Unmanned Aerial Vehicles (UAV) images which include distortion, illumination variance, and large rotation angles are facing multiple challenges of image orientation and image processing. In this paper, a robust and convenient photogrammetric approach is proposed for processing low-altitude UAV images, involving a strip management method to automatically build a standardized regional aerial triangle (AT) network, a parallel inner orientation algorithm, a ground control points (GCPs) predicting method, and an improved Scale Invariant Feature Transform (SIFT) method to produce large number of evenly distributed reliable tie points for bundle adjustment (BA). A multi-view matching approach is improved to produce Digital Surface Models (DSM) and Digital Orthophoto Maps (DOM) for 3D visualization. Experimental results show that the proposed approach is robust and feasible for photogrammetric processing of low-altitude UAV images and 3D visualization of products.

Feature-Based Laser Scan Matching and Its Application for Indoor Mapping

Scan matching, an approach to recover the relative position and orientation of two laser scans, is a very important technique for indoor positioning and indoor modeling. The iterative closest point (ICP) algorithm and its variants are the most well-known techniques for such a problem. However, ICP algorithms rely highly on the initial guess of the relative transformation, which will reduce its power for practical applications. In this paper, an initial-free 2D laser scan matching method based on point and line features is proposed. We carefully design a framework for the detection of point and line feature correspondences. First, distinct feature points are detected based on an extended 1D SIFT, and line features are extracted via a modified Split-and-Merge algorithm. In this stage, we also give an effective strategy for discarding unreliable features. The point and line features are then described by a distance histogram; the pairs achieving best matching scores are accepted as potential correct correspondences. The histogram cluster technique is adapted to filter outliers and provide an accurate initial value of the rigid transformation. We also proposed a new relative pose estimation method that is robust to outliers. We use the lq-norm (0 < q < 1) metric in this approach, in contrast to classic optimization methods whose cost function is based on the l2-norm of residuals. Extensive experiments on real data demonstrate that the proposed method is almost as accurate as ICPs and is initial free. We also show that our scan matching method can be integrated into a simultaneous localization and mapping (SLAM) system for indoor mapping.

Robust Feature Matching for Remote Sensing Image Registration Based on

This letter proposes a robust feature matching algorithm for remote sensing images based on lq -estimator. We start with a set of initial matches provided by a feature matching method such as scale-invariant feature transform and then focus on global transformation estimation from contaminated observations and outliers elimination as well. We use an affine model to describe the global transformation and minimize a new cost function based on lq -norm. We apply an augmented Lagrangian function and an alternating direction method of multipliers to solve such a nonconvex and nonsmooth optimization problem. Extensive experiments on real remote sensing data demonstrate that the proposed method is effective, efficient, and robust. Our method outperforms state-of-the-art methods and can easily handle situations with up to 90% outliers. In addition, the proposed method is much faster than RANSAC.

L_{q}

A multiple terrestrial laser scanner (TLS) integration approach is proposed for the fine surveying and 3D modeling of ancient wooden architecture in an ancient building complex of Wudang Mountains, which is located in very steep surroundings making it difficult to access. Three-level TLS with a scalable measurement distance and accuracy is presented for data collection to compensate for data missed because of mutual sheltering and scanning view limitations. A multi-scale data fusion approach is proposed for data registration and filtering of the different scales and separated 3D data. A point projection algorithm together with point cloud slice tools is designed for fine surveying to generate all types of architecture maps, such as plan drawings, facade drawings, section drawings, and doors and windows drawings. The section drawings together with slicing point cloud are presented for the deformation analysis of the building structure. Along with fine drawings and laser scanning data, the 3D models of the ancient architecture components are built for digital management and visualization. Results show that the proposed approach can achieve fine surveying and 3D documentation of the ancient architecture within 3 mm accuracy. In addition, the defects of scanning view and mutual sheltering can overcome to obtain the complete and exact structure in detail.

-Estimator

Shadows, which are cast by clouds, trees, and buildings, degrade the accuracy of many tasks in remote sensing, such as image classification, change detection, object recognition, etc. In this paper, we address the problem of shadow detection for complex scenes. Unlike traditional methods which only use pixel information, our method joins model and observation cues. Firstly, we improve the bright channel prior (BCP) to model and extract the occlusion map in an image. Then, we combine the model-based result with observation cues (i.e., pixel values, luminance, and chromaticity properties) to refine the shadow mask. Our method is suitable for both natural images and satellite images. We evaluate the proposed approach from both qualitative and quantitative aspects on four datasets. The results demonstrate the power of our method. It shows that the proposed method can achieve almost 85% F-measure accuracy both on natural images and remote sensing images, which is much better than the compared state-of-the-art methods.

Fine Surveying and 3D Modeling Approach for Wooden Ancient Architecture via Multiple Laser Scanner Integration

ABSTRACT Multispectral (MS) and panchromatic (Pan) image fusion, which is used to obtain both high spatial- and spectral-resolution images, plays an important role in many remote-sensing applications such as environmental monitoring, agriculture, and mineral exploration. This article presents an image fusion framework based on the spatial distribution consistency. First, a YUV transform is adopted to separate the luminance component from the colour components of the original MS image. Then, the relationships between the ideal high-resolution multispectral (HRMS) colour components and the Pan band are established based on the spatial distribution consistency, and finally an inverse transform is employed to obtain the fused image. In this article, two types of relationship models are presented. The first model stems from the physical meaning of the assumption and uses a local linear model to describe it. The second model directly uses its algebraic meaning to design the objective cost function and obtains the global optimal solution. The proposed two models are compared with 15 other widely used methods on six real remote-sensing image data sets. Experimental results show that the proposed method outperforms the compared state-of-the-art approaches.

Joint Model and Observation Cues for Single-Image Shadow Detection

ABSTRACT Automatic road extraction from remotely sensed images is an important and challenging task. This article proposes an unsupervised road detection method based on a Gaussian mixture model and object-based features. Our approach has five major stages, i.e. superpixel segmentation, feature description, homogeneous region merging, clustering via the Gaussian mixture model, and outlier filtering. In the third step, we present a graph-based region merging algorithm, in which the nodes of the graph are superpixels and edges are the similarities of intensity, colour, and texture. We also define two shape features, called deviation of parallelism (DoP) and narrow rate (NR), to automatically recognize road layer and filter outliers in the last step. We evaluated the proposed method on a variety of datasets, in which the Vaihingen dataset from the International Society for Photogrammetry and Remote Sensing Test Project is also included. Results demonstrate the power of our approach compared with some state-of-the-art methods.

Multispectral and panchromatic image fusion based on spatial consistency

In the past few years, for its lower-cost, safer and high-resolution images, unmanned aerial vehicles (UAVs) demonstrated great potential for photogrammetric measurements in numerous application fields. Nevertheless, these images are often affected by large rotation, big viewpoint change as well as small overlaps, in which case traditional procedure are not able to orientate images or generate reliable Digital Generation Models (DSM). This paper introduces the whole procedure of the DSM generation, which comprehensively utilizes advantage of both computer vision and multi-image matching algorithms in extracting points and generating a dense DSM. Experiment shows that, based on this procedure, it can quickly extract points from the high-resolution images acquired by UAVs with high location accuracy.

Unsupervised road extraction via a Gaussian mixture model with object-based features

In the past few years, unmanned aerial vehicles (UAVs) demonstrated their great potential for photogrammetric measurements in a lot of application fields because its less expensive, safer and higher resolution images. Nevertheless, their images are often affected by large rotation, big view-point change and small overlaps. In this paper, we present a novel approach for reliable Digital Surface Models (DSM) generation, which is designed to operate on high-resolution, wide-baseline UAV image sets and compute dense 3D point clouds efficiently. It is implemented as a procedure including the four steps of match, expand, filter and reconstruction, starting from a sparse set of matched difference-of-Gaussian (DoG) keypoints, forming a triangulation on it, then expanding per-pixel under local parallax continuity, using visibility constraints to filter false matches, finally generating the DSM. Experiments are conducted to demonstrate the effectiveness and accuracy of our approach and to show that state-of-the-art performance can be achieved with significant acceleration.

An approach of DSM generation from multi-view images acquired by UAVs

High-precision indoor three-dimensional maps are a prerequisite for building information models, indoor location-based services, etc., but the indoor mapping solution is still in the stage of technological experiment and application scenario development. In this paper, indoor mapping equipment integrating a three-axis laser scanner and panoramic camera is designed, and the corresponding workflow and critical technologies are described. First, hardware design and software for controlling the operations and calibration of the spatial relationship between sensors are completed. Then, the trajectory of the carrier is evaluated by a simultaneous location and mapping framework, which includes reckoning of the real-time position and attitude of the carrier by a filter fusing the horizontally placed laser scanner data and inertial measurement data, as well as the global optimization by a closed-loop adjustment using a graph optimization algorithm. Finally, the 3D point clouds and panoramic images of the scene are reconstructed from two tilt-mounted laser scanners and the panoramic camera by synchronization of the position and attitude of the carrier. The experiment was carried out in a five-story library using the proposed prototype system; the results demonstrate accuracies of up to 3 cm for 2D maps, and up to 5 cm for 3D maps, and the produced point clouds and panoramic images can be utilized for modeling and further works related to large-scale indoor scenes. Therefore, the proposed system is an efficient and accurate solution for indoor 3D mapping.