2019 Vol. 44, No. 2
Comprehensive zone correction is a new type of differential corrections for BeiDou wide area augmentation system. As broadcasted together with the equivalent satellite clock and orbit corrections by BDS satellites, they enable user decimeter-level real-time positioning capability using the carrier-phase observations. In this paper, we give a brief introduction of comprehensive zone corrections, and the function model of precise point positioning (PPP) for dual-and single-frequency users using the comprehensive zone corrections. Tracking data of 30 stations in mainland China are used to evaluate the PPP performance, including convergence time, positioning accuracy and its relation with the user's distance from the zone center. Results show that the dual-frequency PPP convergences to 0.5 m in 25 minutes and the positioning accuracy are 0.15 m in horizontal and 0.2 m in vertical, respectively. As for single frequency PPP, the positioning accuracy convergences to 0.8 m in 20 minutes, while the positioning accuracy is 0.3 m in horizontal and 0.5 m in vertical. We conclude that the BDS PPP accuracy using the broadcasted wide area differential corrections reaches decimeter level within the distance of 1 000 km around zone center, and the accuracy becomes slightly worse with the user's distance from the zone center increasing.
Anchoring security is very important to the icebreaker Xuelong during the voyages of polar expeditions. To analyze the feature of anchoring status, we built a decision tree model to recognize the anchoring pattern from the historical trajectory data from the 27th Chinese National Antarctic Research Expedition. On the basis of the recognition model, we retrieve the anchor position during the yawing period using the least squares principle. The fitting anchor position illustrates that the drag-anchor phenomenon occurred when Xuelong was in Antarctic Zhongshan anchorage on February 14, 2011. Combining with the meteorological data, we discuss the cause of drag-anchor and verify the feasibility of near-real-time anchor monitoring by using the fitting anchor position method.
Precise point positioning, as a method for high precision time comparison, has the defects of long latency, poor real-time performance and heavy reliance on precise ephemeris and clock products. A new GPS multi-stations networking for real-time time comparison algorithm is proposed. Redundant observations are obviously increased by multi-stations simultaneous surveying. The clock errors of receivers and satellites are simultaneously estimated with the inputs of IGS ultra-rapid ephemeris. Experiments show that, the time difference between proposed algorithm in this paper and IGS final clock products is less than 0.3 ns, and the daily frequency stability of the alignment results is better than 2.5×10-15.
For the application requirement of route planning and navigation in indoor Location-based-service, we propose a traversable region model based on grid space and its automatic extraction algorithm. Firstly, this paper introduces adjoin grid and traversing grids based on grid model, expounds the basic principle of traversable region model through specific example. Then, according to the characteristics of indoor map data, through indoor grid model initialization, traversable region preliminary extraction and adjoin region merge, it designs traversable region automatic extraction algorithm. Finally, taking Xidan Joy City first floor indoor map data as an example, we perform different grid scale traversable region automatic extraction and route planning experiment. Results show that the automatic extraction algorithm has preferable applicability to deal with complex indoor environment which has barrier in the corridor and so on. Compared with network model, the route planning results of traversable region model are more conformed to the route walking feature of complex indoor environment.
Quality control is described by internal and external reliability according to the Delft philosophy. Internal reliability describes the ability to find biases in observational data and is represented accordingly by the minimal detectable bias (MDB). The MDBs formulas of four popular outliers detection methods include data snooping, quasi-accurate detection of gross errors(QUAD), partly least squares (PLS) and the method of simultaneous locating and evaluating multiple gross errors(LEGE) are derived with the given significance level α0 and given power of test β0. The numerical example reveals that MDBs are same value when the observations have the same weight. If the clean data are not equally weighted, the MDBs of data snooping are equal to that of QUAD and PLS, but all are slightly smaller than the MDBs of LEGE. If the clean data are correlated observations, the MDBs of QUAD are equal to that of PLS, but both are larger than that of data snooping and smaller than that of LEGE.
To detect clock jump and gross error in data of satellite clock, a two-step method for error detection of satellite clock data is proposed by combining the single difference algorithm of epochs and quasi-accurate cetection method (QUAD). In the first step, a single difference algorithm of epochs is used to detect suspected clock jump and eliminate the gross error points according to the different performance characteristic of clock jump and gross error in the sequence of single difference of epochs; and in the second step, QUAD is used to detect gross error precisely for the data by segment according to the suspected clock jump. The detecting process is set forth and the performance is analyzed by using simulating data, and the algorithm is verified by practical data, both examples show that using only single difference algorithm of epochs can effectively detect suspected clock jump points and partial error, while using the two-step method can effectively avoid clock jump influence, and can accurately detect small gross errors, which is close to two times the background noise's standard deviation.
In order to improve the precision of the innermost area effects in altimetry gravity computed by the inverse Vening-Meinesz formula, deflections of the vertical are expressed as bi-quadratic polynomials regarding the innermost area as a rectangular one, and the formulas to calculate gravity anomaly of this area are derived after the non-singular transformation is introduced. A practical calculation is done based on deflections of the vertical data with a resolution of in the low latitude area. The results indicate that the maximal difference between the contributions of the innermost area including four grids calculated by traditional formulas and this paper's formulas is greater than 1 mGal. The formulas derived in this paper can provide theoretical basis for the innermost area effects in altimetry gravity with high precision.
Among existing theories on precision estimation of nonlinear adjustment, the simulation number of Monte Carlo method generally is chosen subjectively and its result also cannot be controlled directly. Besides those, the biases of parameter estimates, corrections of observations and the estimate of variance of unit weight are not taken into consideration simultaneously. The adaptive Monte Carlo method is combined with precision estimation of nonlinear adjustment for solving problems given above in this paper. By calculating biases of estimates and covariances matrix of parameter estimates, the complete process of precision estimation based on adaptive Monte Carlo is given. With the help of the term of antithetic variates, the antithetic and adaptive Monte Carlo algorithm is proposed for biases of parameter estimates. Results from two examples of straight line fitting model and ellipse fitting model show that the adaptive Monte Carlo method in this paper can obtain the stable and reasonable effects for precision estimation of nonlinear adjustment with extensive applicability, the antithetic and adaptive Monte Carlo in this paper is better at convergence and computational efficiency for calculating biases of parameter estimates.
Considering the non-random elements and the same random elements in different locations of coefficient matrix, the partial errors-in-variables(PEIV) model is stricter than general EIV model. However, the existing weighted total least squares (WTLS) algorithm for PEIV model requires more iteration so that the computation efficiency is reduced. Making using of statistical properties of observations errors and coefficient matrix error, this paper deduces a new computational formula of WTLS estimation for PEIV model based on Fisher-Score algorithm. The numerical results show that the Fisher-Score algorithm takes less iterations and the computation efficiency is enormously promoted.
According to the issue that zero-difference precise point positioning(PPP) augmentation information cannot be directly applied to real-time kinematic(RTK) relative positioning, this paper studies the method of obtaining the zero-difference correction information with PPP ambiguity resolution based on additional coordinate constraint in the reference station, establishes the model of equivalence transformation from zero-difference augmentation information to virtual observation data, and focuses on difference application of state space representation information in equivalence transformation. Experiment designed with real GNSS observation data proves correctness and availability of algorithm according to RTK ambiguity with partial ambiguity fixing technology. The results show that virtual zero-baseline can deliver a good accuracy positioning as same as the network RTK, thereby enabling high unification of regional augmentation system in terms of zone-difference and difference modes.
A double truncated singular value estimation based on signal-to-noise ratio test is proposed by combining the measurement results of the harm degree that complex common linear exerts on parameter estimation and the truncated singular value decomposition(TSVD) estimation. Based on the signal to noise ratio(SNR) test, all the parameters are divided into two parts, according to the estimated SNR value of each parameter's least square estimation. The TSVD estimation of these two parts is truncated at different intensities. We choose a smaller truncated parameter for parameters which suffer more from complex common linear, and a bigger truncated parameter for parameters which suffer less from complex common linear, thus minimize the deviation and reduce the variance of parameter estimation effectively. This new method is applied to the simulation example of the GEO satellite orbit determination. The experimental results show that the new method is more accurate.
In UWB indoor navigation, the accuracy of navigation resolution is greatly affected by non line of sight(NLOS) ranging error, and low filtering precision is influenced by uncertain system noise. To solve these problems, an improved robust Kalman filtering based on innovation is proposed and applied in ultra wideband(UWB) indoor navigation. On the foundation of the linear UWB indoor navigation model, the new method uses single innovation values to construct the matrix with robust factors and eliminate the influence of NLOS ranging error. Meanwhile, the new method does real-time estimation and corrects the system noise covariance matrix. Experimental result verifies the effectiveness of the new method. It is shown that the new method can not only effectively eliminate the influence of NLOS ranging error on navigation resolution, but also can further improve the filter precision and reliability in indoor navigation.
Based on the systematic collection of the GPS observations in China:Fujian, China:Taiwan and Philippine:Luzon, this paper obtains the current crustal horizontal velocity field of the arc-continent collision zone. Targeting the specific areas whose super-long span is more than 1000 km, this paper builds the least squares collocation model on the basis of ellipsoidal coordinates, uniformly calculating current GPS strain distribution of this collision zone. The results show that china:Taiwan province has the strongest crustal deformation, followed by Philippine:Luzon and China:Fujian regions respectively. The Philippine sea plate at the east Taiwan, China collides intensively with the Eurasia plate, and the collision is severely consumed through the central mountain's uplifting, the thrust napping and the tectonic escaping. The overall deformation characteristics is closely related to the structural variation of the Philippine sea plate. While the relative deformation in Luzon region is mainly involved with the relative sliding of the Philippine fault.
To improve the accuracy of unmanned helicopter, geometry calibration is compulsory. Direct geo-reference experiment is first carried out to verify system's accuracy without any calibration; and then the free network bundle adjustment experiment is undertaken to verify the positioning accuracy of the system without ground control points; finally, a self calibration bundle adjustment is undertaken to calibrate and compensate the camera distortion, interior elements changes, IMU alignment angle. Experimental results show that the unmanned helicopter has high positioning accuracy. The accuracy of direct geo-reference can meet the requirements of emergency mapping, and can be aided by a few ground control points, the positioning accuracy is much better than that of 1:500 mapping requirements, which can be used for conventional surveying and mapping production. After system error compensation, the direct geo-reference accuracy has improved significantly:the planar accuracy has improved by 46.37% and the height accuracy has improved by 62.49%. The flight test of Songshan calibration field shows that the method of calibration and compensation is correct and effective, and the unmanned helicopter has a good prospect for surveying and mapping applications.
Nowadays, the texture mapping for 3D indoor model mostly relies on heavy artificial operation. Its inefficiency and difficulty in realizing texture updating after the change of indoor environment leads to a lot of out-dated and unreal texture in time. To solve this problem, we propose a new method for updating 3D indoor model texture automatically. Using a single realistic image from smartphone, we firstly estimate the spatial layout of the room in the image based on 3D indoor scene understanding. Then, according to phone's direction and position, we find the triangular patches of the targeted wall in the 3D geometric model. At the end, the (u, v) coordinates of the image are calculated by matching the convex points of the layout box with the corresponding patches of the 3D model. The conducted experiments demonstrate that the proposed method can automatically map a single image onto the 3D indoor model, which is meaningful for the augmented reality in the indoor space.
This paper presents an effective algorithm to extract key frames from video images based on the correlation coefficients derived from the overlapping regions of adjacent images. Firstly, the polynomial model is employed to fit the change trend of correlation coefficients. Secondly, the positions of key frames are located and extracted rapidly and correctly via the high correlation of overlapping regions of adjacent key frame images. Finally, the time-effectiveness of extracting key frames based on the degree of overlap is tested in a stance. The result shows that the proposed algorithm takes only near 1/4 of time consumed by single computing of the degree of overlap, with an accuracy up to 88% and it can achieve higher accuracy for extracting key frames with different requires of degree of overlap between video images. The panchromatic image generated from the mosaic of extracted key video frames shows no apparent missing of scenes, and the new algorithm demonstrates high reliability and transferability.
Traditional camera calibration methods are limited by high-precision three-dimensional inspection tinker or calibration template with known space structure. Aiming at this problem, this paper puts forward a kind of camera calibration method without control information on the basis of the analysis of characteristics of smartphone imaging. This method includes two steps:Firstly, we collect images data according to 3×3 mode, select a reference image to take relative orientation with other images in order to build unit model, and build free net model through the proportional coefficient after model connection, then coordinates of projective centers and object space points will be all taken into one coordinate system of reference image; secondly, taking the base image selected as reference and considering additional parameter model, orientation elements of images relative to reference image and additional parameters will be gotten after self-calibration bundle block adjustment. Experimental results show that this calibration method is effective and can significantly improve the accuracy of measurement.
Pedestrian detection is a hot topic and difficult problem in areas like computer vision and intelligent traffic. This paper researches on pedestrian detection in complex indoor space using depth sensors. In recent years, target detection methods based on RGB-Depth(RGBD) data mainly include background learning method and feature detecting operator method. However, the former method depends on the background knowledge of first tens of frames, and the number of frames decides the final detection accuracy. The latter takes plenty of time for computing, and being lacks of training samples may influence the detection result. Thus, this paper analyzes the complex scene features and integrates the color and depth information, and proposes a RGBD+ViBe (visual background extractor) background elimination method. The experiment results indicate that the detection accuracy of the proposed method is higher than the methods which only consider color or depth information and the RGBD+MoG method in foreground extraction.
This paper focuses on the detection of gross error in airborne SAR image coordinates during the process of orientation parameter calculation and stereo positioning. Firstly, based on the range Doppler model, a preliminary discussion is conducted to investigate the impact of image coordinate gross error on the error equations. Secondly, after analyzing the necessity and difficulty of gross detection, based on the quasi-accurate detection, a specific calculation scheme and strategy for orientation parameter calculation is proposed. For the first time, quasi-accurate detection is applied to SAR image orientation. Finally, a series of experiments with both simulated data and real airborne SAR images are carried out to validate the proposed scheme and strategy. The experimental results show that the proposed scheme in this paper is able to detect and calculate the gross errors in image coordinates. Compared to the classical least-square estimation usually used in image orientation, a higher accuracy of both orientation parameters and the subsequent stereo positioning are achieved. The research in this paper is particularly meaningful to improve the quality of the SAR data affected by gross errors in situations where control points are not enough.
Ground-based interferometric radar can realize small regional-scale continuous deformation monitoring with high-accuracy. To investigate the deformation detection capability and accuracy, a set of accuracy verification platform and system are established in this paper. The corner reflectors are arranged at different positions on the roof and the stepping platforms are used to control the corner reflectors and simulate deformation in different values. The deformation corner reflectors, stable points and the surface of the roof constitute a small two-dimensional deformation field. We use the IBIS-L system to finish the deformation detection and analysis. The experimental results show that when the millimeter level deformation occurs in the corner reflectors, the average accuracy of deformation detection of the ground-based interferometric radar IBIS-L system is 0.27 mm, while the sub millimeter level deformation occurs the corner reflectors, its average accuracy is 0.11 mm. This system can realize the sub millimeter level deformation detection in small regional-scale. For the small and slow change, this system has better deformation detection ability and reliability.
Aimed at the conflict between load balancing for user access services and sequentially reading for disk storage services, a user-driving storage and organization strategy for spatial data is proposed, which takes a comprehensive consideration of the strategy of data placement and load balance (CSDL). This scheme mines the users' behaviors and computes the correlations among all data so as to distribute and store the popular data into different storage nodes to realize load balancing. Then, the concurrency degree is also computed among the data stored in the same storage node and which can be used to store some data in contiguous disk space to realize continuous reading. The CSDL method proposed in this paper tries to organize spatial data storage from two aspects of load balancing and lower disk efficiency at the same time, so as to improve the service efficiency of GIS. Experimental results show that our scheme improves the performance of average request response time by 45.2%-245.3% and also improves the performance of load balance degree by 0.5%-440.9%, which can meet the requirements of large scale distributed environments.
Different from the conventional address word segmentation model, which relies on the city address dictionary or the rule set, this paper proposes a word segmentation method which does not depend on the address dictionary but based on massive address data mining. This method combines the statistic rules to calculate the distribution of the address elements in the address dataset, excavates the suffix points and the drop points of the address elements in the address data. The method constructs the statistical decision tree based on their relative position relations to extract the address elements, uses the investigation data of building address in Shenzhen to verify and to make a useful supplement to the current gazetteers.
The road selection is a cartographic generalization operation to retain more important parts, or to omit the less important parts, of the linear features representing road network according to the cartographic scale requirement. Many researchers have paid much attention to the approaches to road network selection. This paper gives a comparative analysis of two selection strategies from conceptual and operational level. One strategy is to update the importance of the retained roads after eliminating one road and the other is not to update. The stroke ordering method is used as the road selection method, and length, connectivity, closeness, and betweenness are used to determine the importance of individual strokes. The selection results are evaluated by quantitative measures (including similarity, commission error and omission error), and qualitative visual inspection. The road network of Shenzhen city is used in the experiments. Theoretically, the update after dimination strategy is better than non-update strategy. Empirically, quantitative results shows that the non-update strategy is superior, while qualitative inspection shows that each strategy has its own advantages and disadvantages.