Pose From Orthography And Scaling With Iterations Research Articles

Visual Navigation Method for Carrier-based Aircraft Landing

Carrier landing of fixed-wing aircraft is the most difficult and challenging task in the aerial field. Precise and reliable landing systems are the key to carrier landing. Visual navigation has the advantages of high precision of the relative navigation. Hence, it has become a promising method for landing systems. However, tracking the targets continuously and extracting the coordinates of the targets accurately are challenges for visual navigation due to the complex environment and dramatic distance changes during carrier landing. Therefore, a visual navigation method utilizing an airborne medium-wave infrared camera to track four infrared targets installed on the carrier, which can operate in low visibility weather conditions, is presented in this article. A segmentation detection method was adopted for tracking the targets. The kernel correlation filter was used to detect the targets when the aircraft was far from the carrier stern, which could quickly track targets with relatively constant dimensions. A convolutional neural network classifier based on transfer learning was used to detect the targets when nearing the carrier, thereby reducing the risk of losing targets due to severe dimensional changes. Finally, the relative position and attitude information could be obtained by the POSIT (pose from orthography and scaling with iterations) pose estimation algorithm. On-vehicle experiments were performed, and the results verified the effectiveness of the method.

Accuracy of 3D reproduction of natural head position using three different manual reorientation methods compared to 3D software

PurposeThe aim of this study was to analyze the positional differences of three-dimensional (3D) natural head positions (NHPs) reproduced by three different manual reorientation methods without special software by the Pose from Orthography and Scaling with ITerations (POSIT) method. MethodsFive ceramic markers were attached to each of 12 patients’ faces, and frontal and lateral photographs in the NHP and 3D computed tomography (CT) were taken. The 3D surface model was reoriented for the NHP reproduction by four different methods: the POSIT method (standard method), the location of the markers (A), the soft tissue landmarks (B) on the photographs, and manual correction without photographs (C). On each 3D surface model, the location of the skull was evaluated three-dimensionally. ResultsDifferences between reproduced NHPs in each of the four different methods were statistically significant (p < 0.0001). Compared to the POSIT method, the accuracy of the other reproducing methods was lower. The A and B methods showed a similar accuracy to each other, while the C method presented the most inaccurate NHP. ConclusionIf 3D NHP reproduction using special software is impossible, reproducing NHP with photographs may be used as an alternative method, but its application should be clinically limited.

Three-dimensional natural head position reproduction using a single facial photograph based on the POSIT method

We developed a new method to record and reproduce the three-dimensional natural head position (NHP) from a single photograph of a patient's face using a pose from orthography and scaling with iterations (POSIT) algorithm. We attached 4-mm spherical ceramic markers to the patient's face as feature points. A frontal photograph of the patient's NHP was taken using an ordinary digital camera parallel to the global horizon. Computed tomography (CT) was then performed on the patient with the markers. The ceramic marker positions were determined in the 2D image and corresponded to points in the 3D model. The 3D rotation matrix determined using the feature points via the POSIT method was applied to the CT model to reproduce the NHP. A skull phantom was used to evaluate the accuracy and reproducibility of the developed method. The degree difference (°) between the true and POSIT orientations in the roll, pitch, and yaw directions was quantified as the error. The mean accuracy was −0.04 ± 0.15°, −0.17 ± 0.50°, and −0.02 ± 0.37° in the roll, pitch, and yaw directions, respectively. The method developed was highly reproducible during intra-observer and inter-observer variation analyses. The accuracy of the method was clinically acceptable, and the procedure was time- and cost-effective. This method is accurate and inexpensive; additionally, it does not affect the patient's lip position, and we expect it to be routinely used during orthognathic surgery.

48.3: Viewer's Eye Position Estimation using Single Camera

AbstractIn this paper, we propose a system that can estimate viewer's eye position using single camera in an autostereoscopic 3D display, which adopts Active Appearance Model (AAM), Elastic Bunch Graph Matching (EBGM) and Pose from Orthography and Scaling with ITerations (POSIT) algorithms for the eye position calculation. AAM is one of the most prominent facial feature detection algorithms for estimating viewer's eye position. However, this algorithm also has an issue in accuracy. EBGM is also known to be a very successful algorithm for the detection of viewer's facial feature points. Also, POSIT is known to calculate 3D pose from 2D image. In this paper, we propose a way to improve the accuracy of viewer's eye position by applying algorithms, AAM for initial estimation of facial features, EBGM for accurate facial feature position calculation and POSIT for calculation 3D facial position. The system estimates 3D facial position from extracted facial feature points to calculate accurate eye position. To calculate accurate eye position, we adopt POSIT algorithm when the viewer's face is rotated. It is verified through experiments that the newly proposed method performs better than existing method that uses single camera.

A 3-degree of freedom binary search pose estimation technique

This paper describes a novel, binary search pose estimation (BSPE) technique for poses constrained to 3 degrees of freedom (DOF). The technique requires three fiduciary marker points and operates by minimising the angular DOF through a binary search driven algorithm. The technique is less computationally intensive than the standard closed form solutions and quickly converges to a solution which can be verified if additional fiduciary points are available. Evaluation for the technique is presented through a comparison to pose from orthography and scaling with iterations (POSIT) using synthetic data and a real-world application, where the technique is used for pose estimation of a mobile robot with respect to a motor vehicle for inspection purposes.

Model-based object pose in 25 lines of code

In this paper, we describe a method for finding the pose of an object from a single image. We assume that we can detect and match in the image four or more noncoplanar feature points of the object, and that we know their relative geometry on the object. The method combines two algorithms; the first algorithm,POS (Pose from Orthography and Scaling) approximates the perspective projection with a scaled orthographic projection and finds the rotation matrix and the translation vector of the object by solving a linear system; the second algorithm,POSIT (POS with ITerations), uses in its iteration loop the approximate pose found by POS in order to compute better scaled orthographic projections of the feature points, then applies POS to these projections instead of the original image projections. POSIT converges to accurate pose measurements in a few iterations. POSIT can be used with many feature points at once for added insensitivity to measurement errors and image noise. Compared to classic approaches making use of Newton's method, POSIT does not require starting from an initial guess, and computes the pose using an order of magnitude fewer floating point operations; it may therefore be a useful alternative for real-time operation. When speed is not an issue, POSIT can be written in 25 lines or less in Mathematica; the code is provided in an Appendix.