Oblique View Angle Research Articles

High Throughput Image Analytics Using Hough Transformation for Wheat Phenotyping

HighlightsRGB-based vegetation and leaf area indexes using a smartphone camera.Geo-rectification of skewed images via row detection using Hough Transformation.Open-source software to automate the image stitching and plot-level phenotypic metrics extraction. Abstract. An agricultural field is not always accessible for plant phenotyping with existing mobile platforms due to the limited space and regulated aviation area. Smartphone-triggered ground images were collected on a wheat field that has a limited access to monitor growth conditions of four wheat varieties: Shinyoung (SY), Joseong (JS), Taewoo (TW), and Cheongwoo (CW). For field mapping during the growing season, six sets of the raw images were acquired by a smartphone in an oblique view angle and processed to transform into nadir view images. Algorithms were developed to process the raw tile images for geometric rectification via row detection using Hough Transformation. Stitching software was developed to automate the high throughput image analytics of the skewed tile images into a stitched field image through deskewing, row alignment, overlap trimming, and resizing. Plot-level metrics were extracted to analyze plant growth of the wheat varieties using a gridding method for vegetation and leaf area indexes. The processed images resulted in the successful transformation and consistency of algorithms on image alignment and stitching. Plot-level analysis indicated that SY variety performed superior to the other varieties in both vegetation and leaf area indexes and was significantly different in the canopy coverage from the least performed TW variety. The image analytic methods developed in the study offer a flexible solution to stitch and align tile images by a hand-held camera in both oblique and nadir view via user-friendly interface software for high through plant phenotyping and can be adapted to other stationary or mobile imaging platforms in greenhouse and fields. Keywords: Calibration, Image processing, Phenotyping, Python, Software, Stitching.

Measurement of land surface temperature from oblique angle airborne thermal camera observations

ABSTRACTThis paper presents a novel airborne remote sensing method using thermal imaging to directly georeference and calculate Earth surface temperature with a high spatiotemporal resolution. A tethered balloon is used to elevate an uncooled thermal camera in the field. When deployed, images with oblique view angles of the surrounding Earth surface are collected. Images recorded from a field environmental monitoring campaign in a northern Canadian mining facility are processed with open source software, and it is shown that they successfully represent the diurnal and spatial surface temperature variations within the facility. Furthermore, in comparison to MODerate resolution Imaging Spectroradiometer (MODIS) satellite images, the approach results in a median absolute error of 0.64 K, with a bias and Root Mean Square Error (RMSE) of 0.5 K and 5.45 K, respectively.

Visual Pose Estimation of Rescue Unmanned Surface Vehicle From Unmanned Aerial System.

This article addresses the problem of how to visually estimate the pose of a rescue unmanned surface vehicle (USV) using an unmanned aerial system (UAS) in marine mass casualty events. A UAS visually navigating the USV can help solve problems with teleoperation and manpower requirements. The solution has to estimate full pose (both position and orientation) and has to work in an outdoor environment from oblique view angle (up to 85° from nadir) at large distances (180 m) in real-time (5 Hz) and assume both moving UAS (up to 22 m s−1) and moving object (up to 10 m s−1). None of the 58 reviewed studies satisfied all those requirements. This article presents two algorithms for visual position estimation using the object's hue (thresholding and histogramming) and four techniques for visual orientation estimation using the object's shape while satisfying those requirements. Four physical experiments were performed to validate the feasibility and compare the thresholding and histogramming algorithms. The histogramming had statistically significantly lower position estimation error compared to thresholding for all four trials (p-value ranged from ~0 to 8.23263 × 10−29), but it only had statistically significantly lower orientation estimation error for two of the trials (p-values 3.51852 × 10−39 and 1.32762 × 10−46). The mean position estimation error ranged from 7 to 43 px while the mean orientation estimation error ranged from 0.134 to 0.480 rad. The histogramming algorithm demonstrated feasibility for variations in environmental conditions and physical settings while requiring fewer parameters than thresholding. However, three problems were identified. The orientation estimation error was quite large for both algorithms, both algorithms required manual tuning before each trial, and both algorithms were not robust enough to recover from significant changes in illumination conditions. To reduce the orientation estimation error, inverse perspective warping will be necessary to reduce the perspective distortion. To eliminate the necessity for tuning and increase the robustness, a machine learning approach to pose estimation might ultimately be a better solution.

SU‐F‐T‐11: Scintillator Based Quality Assurance Device for HDR Brachytherapy

Purpose:To build a test device for HDR afterloaders capable of checking source positions, times at positions and estimate the activity of the source.Methods:A catheter is taped on a plastic scintillation sheet. When a source travels through the catheter, the scintillator sheet lights up around the source. The sheet is monitored with a video camera, and records the movement of the light spot. The center of the spot on each image on the video provides the source location, and the time stamps of the images can provide the dwell time the source spend in each location. Finally, the brightness of the light spot is related to the activity of the source. A code was developed for noise removal, calibrate the scale of the image to centimeters, eliminate the distortion caused by the oblique view angle, identifying the boundaries of the light spot, transforming the image into binary and detect and calculate the source motion, positions and times. The images are much less noisy if the camera is shielded. That requires that the light spot is monitored in a mirror, rather than directly. The whole assembly is covered from external light and has a size of approximately 17×35×25cm (H×L×W)Results:A cheap camera in BW mode proved to be sufficient with a plastic scintillator sheet. The best images were resulted by a 3mm thick sheet with ZnS:Ag surface coating. The shielding of the camera decreased the noise, but could not eliminate it. A test run even in noisy condition resulted in approximately 1 mm and 1 sec difference from the planned positions and dwell times. Activity tests are in progress.Conclusion:The proposed method is feasible. It might simplify the monthly QA process of HDR Brachytherapy units.

Optimal Angle of Needle Insertion for Fluoroscopy‐Guided Transforaminal Epidural Injection of L5

Unlike other sites, there is difficulty in performing TFESI at the L5-S1 level because the iliac crest is an obstacle to needle placement. The objective of this study was to identify the optimal angle of fluoroscopy for insertion and advancement of a needle during L5 TEFSI. We conducted an observational study of patients undergoing fluoroscopy-guided L5 TFESI in the prone position. A total of 80 patients (40 men and 40 women) with radiating pain of lower limbs were enrolled. During TFESI, we measured the angle at which the L5 vertebral body forms a rectangular shape and compared men and women. Then, we measured area of safe triangle in tilting angle of fluoroscopy from 15° to 35° and compared men and women. The mean cephalocaudal angle, where the vertebral body takes the shape of a rectangle, was 11.0° in men and 13.9° in women (P=0.007). In men, the triangular area was maximal at 18.3mm² with an oblique view angle of 25°. In women, the area was maximal at 23.6mm² with an oblique view angle of 30°. At an oblique view angle of 30° and 35°, the area was significantly greater in women (P<0.05). When TFESI is performed at the L5 region in the prone position, placement of fluoroscopy at a cephalocaudal angle of 11.0° and an oblique angle of 25° in men and cephalocaudal angle of 13.9° and an oblique angle of 30° in women would be most reasonable.

In-scene atmospheric correction of hyperspectral thermal infrared images with nadir, horizontal, and oblique view angles

Atmospheric corrections for hyperspectral thermal images acquired with nadir, horizontal, and oblique views have typically relied on atmospheric modelling software, such as Moderate Resolution Atmospheric Transmission (MODTRAN), to estimate atmospheric parameters. Data-only corrections, which require only information from the scene, are more versatile and less labour intensive, but do not yet seem to have been applied to horizontal and oblique views. Here, we apply, and modify where necessary, one published data-only algorithm (in-scene atmospheric correction (ISAC)) to nadir, horizontal, and slanted views (The Aerospace Corporation's Spatially Enhanced Broadband Array Spectrograph System (SEBASS) and Telops Inc.'s Hyper-Cam data sets) to assess the applicability for different viewing geometries. We find that it successfully retrieves characteristic mineralogical emissivity spectra in scenes taken from all geometries. We also discuss additional corrections to the Hyper-Cam data set to correct for instrumental artefacts.

Accurate Localization of Metal Electrodes Using Magnetic Resonance Imaging

Purpose : Localization using MRI is difficult due to susceptibility induced artifacts caused by metal electrodes. Here we took an advantage of the B0 pattern induced by the metal electrodes by using an oblique-view imaging method. Materials and Methods : Metal electrode models with various diameters and susceptibilities were simulated to understand the aspect of field distortion. We set localization criteria for a turbo spin-echo (TSE) sequence usingconventional (90� view) and 45�oblique-view imaging method through simulation of images with various resolutions and validated the criteria usingphantom images acquired by a 3.0T clinical MRI system. For a gradient-refocused echo (GRE) sequence, which is relatively more sensitive to field inhomogeneity, we used phase images to find the center of electrode. Results : There was least field inhomogeneity along the 45�line that penetrated the center of the electrode. Therefore, our criteria for the TSE sequence with 45� obliqueview was coincided regardless of susceptibility. And with 45�oblique-view angle images, pixel shifts were bidirectional so we can detect the location of electrodes even in low resolution. For the GRE sequence, the 45�oblique-view anglemethod madethe lines where field polarity changes become coincident to the Cartesian grid so the localization of the center coordinates was more facilitated. Conclusion : We suggested the method for accurate localization of electrode using 45� oblique-view angle imaging. It is expected to be a novelmethodto monitoring an electrophysiological brain study and brain neurosurgery.

Composite rock slope kinematics at the current Randa instability, Switzerland, based on remote sensing and numerical modeling

Kinematic analysis of slope instabilities in brittle rock is crucial for understanding the reaction of the rock mass to external forcing factors. In steep terrain, inaccessibility often limits collection of relevant data and remote sensing techniques must be applied. This is the case at the current Randa rock slope instability in southern Switzerland, where a total volume of about 5–6 million m 3 moves at a rate of up to 30 mm/yr. A large portion of the unstable rock mass is exposed in an 800 m high inaccessible cliff; the main scarp of the May 1991 rock slope failure. Between 2005 and 2007, a comprehensive suite of remote sensing techniques, including photogrammetry, LiDAR, and GB-DInSAR, was combined with 3D geodetic measurements to characterize the rock mass structure and displacement patterns. Photogrammetry and LiDAR data were measured simultaneously from a helicopter using a system allowing for oblique view angles, which provided optimal observations of the steep rock cliff. We used these datasets to map large-scale structures and extract their orientation and minimum persistence, as well as to characterize the 1991 failure scarp. The northern part of the May 1991 failure surface shows a transition from stepped planar sliding at the base, to failed rock bridges in the center, to tensile failure close to the vertical head scarp. Kinematic analysis of the discontinuity sets in the currently moving rock mass shows that both toppling and translational sliding are feasible failure mechanisms. Toppling is more likely for steep faces above 2200 m, whereas translational failure is more likely in the lower portion of the instability. Interpretation of GB-DInSAR displacement maps revealed similar kinematic behavior, and also allowed identification of a basal rupture zone and lateral release plane bounding the instability. Displacement vectors derived from geodetic surveying provided new insights into the 3D kinematic behavior of the instability. All information extracted from different data sets were integrated in a conceptual model, which was then investigated with 2D numerical simulation using the discontinuum code UDEC. The numerical models were able to reproduce the hypothesized kinematic behavior well.

Cloud model evaluation using radiometric measurements from the airborne multiangle imaging spectroradiometer (AirMISR)

Detailed information on cloud properties is needed to rigorously test retrieval algorithms for satellite and ground-based remote sensors. The inherent complexity of clouds makes this information difficult to obtain from observations alone and cloud resolving models (CRMs) are often used to generate synthetic datasets that can be used as proxies for real data. We test the ability of a CRM to reproduce the observed structure of low-level clouds detected by the Earth Observing System (EOS) validation program in north central Oklahoma on March 3, 2000. A three-dimensional radiative transfer model is applied to high-resolution cloud fields generated by the CRM in order to simulate the top of atmosphere radiances. These synthetic radiances are then statistically compared with observations from the airborne Multiangle Imaging SpectroRadiometer (AirMISR), flown on the NASA ER-2 high-altitude aircraft. Simulations match well the blue channel radiance distributions at oblique view angles but overestimate the minimum in reflectance at near-nadir viewing angles. The model reproduces correctly the angular change in the width of radiance distribution, as measured by the standard deviation but the change in skewness of these distributions is captured only qualitatively. The model biases are suggestive of the simulated cloud boundaries being too sharp and the distribution of the vertical liquid water path being too narrow. A power spectrum analysis shows a close agreement between simulations and observations including a break in the scaling properties of the radiance fields around 400 m. The scaling properties of the 1D power spectrum of simulated radiance at smaller scales exhibit directional dependence not seen in corresponding observations.

Quantifying Vegetation Change by Point Sampling Landscape Photography Time Series

Quantitative assessment of vegetation change is often conducted by digitally analyzing time series of aerial or vertical photography. Change analysis conducted using repeated oblique or landscape photography, however, has been limited to qualitative assessments. The purpose of this study was to develop sampling and analysis techniques for using a time series of digitized landscape photography to quantify vegetation change on rangeland landscapes. Digital images were created from black-and-white landscape photographs acquired in 1917, 1962, and 2000 near Whiskey Mountain in the Reynolds Creek Experimental Watershed of southwestern Idaho. Images were spatially registered to each other using control points and a polynomial transformation algorithm. Thirty random pixels along each of 30 random image lines were selected as point samples (n = 900) from within each image. The landscape feature represented in each selected pixel was classified into 1 of 15 cover types. Cover-type classification accuracy for the 2000 image was estimated to be 92.2% based on ground-truth data collected in the field. Classification accuracy was increased to 98.9% by combining rare or poorly separable cover-type classes. Image cover of vegetation cover types was quantified for each photography acquisition date. Changes in image cover of each cover type and direction of cover-type conversions were determined for each intervening time period. Analysis of image cover using repeated landscape photography is constrained by limitations imposed by oblique view angles and variable image quality. Repeat landscape photography, however, can be used to quantitatively assess long-term dynamics of vegetation cover on rangeland landscapes with visually distinct vegetation types. La evaluación cuantitativa del cambio de la vegetación a menudo es conducida por el análisis digital de series de tiempo de fotografías aéreas o verticales. El análisis de cambio usando fotografía oblicua repetida o de paisaje ha estado limitada a evaluaciones cualitativas. El propósito de este estudio fue desarrollar técnicas de muestreo y análisis para usar series de tiempo de fotografías de paisaje digitalizadas para cuantificar el cambio de vegetación en paisajes de pastizal. Se crearon imágenes digitales partir de fotografías de paisaje en blanco y negro adquiridas en 1917, 1962 y 2000 cerca de la montaña Whiskey en la Cuenca Hidrológica Experimental Reynolds Creek en el suroeste de Idaho. Las imágenes fueron registradas espacialmente una de otra usando puntos de control y un algoritmo de transformación polinomial. Como puntos de muestreo dentro de cada imagen se seleccionaron aleatoriamente 30 pixeles a lo largo de cada una de 30 líneas de imagen aleatorias (n = 900). La característica del paisaje representada en cada píxel seleccionado fue clasificada dentro de 1 de 15 tipos de cobertura. La exactitud de la clasificación por tipo de cobertura para las 2000 imágenes fue estimada en 92.2%, basada en datos verdaderos colectados en campo. La exactitud de la clasificación aumentó a 98.9% al combinar clases de tipo de cobertura raras o pobremente separables. La imagen de cobertura de los tipos de cobertura de vegetación fueron cuantificados para cada fecha de adquisición de las fotografías. Los cambios en la imagen de cobertura de cada tipo de cobertura y la dirección de las conversiones de tipo de cobertura fueron determinadas en cada período de tiempo que intervino. El análisis de las imágenes de cobertura usando la fotografía repetida del paisaje esta restringido por limitaciones impuestas por el ángulos de vista oblicuos y la calidad variable de la imagen. Sin embargo, la fotografía repetida de paisaje puede ser usada para evaluar cuantitativamente las dinámicas a largo plazo de la cobertura de vegetación de los paisajes de pastizal con tipos de vegetación visualmente distintos.

Quantifying Vegetation Change by Point Sampling Landscape Photography Time Series

Quantitative assessment of vegetation change is often conducted by digitally analyzing time series of aerial or vertical photography. Change analysis conducted using repeated oblique or landscape photography, however, has been limited to qualitative assessments. The purpose of this study was to develop sampling and analysis techniques for using a time series of digitized landscape photography to quantify vegetation change on rangeland landscapes. Digital images were created from black-and-white landscape photographs acquired in 1917, 1962, and 2000 near Whiskey Mountain in the Reynolds Creek Experimental Watershed of southwestern Idaho. Images were spatially registered to each other using control points and a polynomial transformation algorithm. Thirty random pixels along each of 30 random image lines were selected as point samples (n = 900) from within each image. The landscape feature represented in each selected pixel was classified into 1 of 15 cover types. Cover-type classification accuracy for the 2000 image was estimated to be 92.2% based on ground-truth data collected in the field. Classification accuracy was increased to 98.9% by combining rare or poorly separable cover-type classes. Image cover of vegetation cover types was quantified for each photography acquisition date. Changes in image cover of each cover type and direction of cover-type conversions were determined for each intervening time period. Analysis of image cover using repeated landscape photography is constrained by limitations imposed by oblique view angles and variable image quality. Repeat landscape photography, however, can be used to quantitatively assess long-term dynamics of vegetation cover on rangeland landscapes with visually distinct vegetation types.

Posterior microendoscopic surgical approach for the degenerative cervical spine

We describe the advantages and clarify the technical key points of a microendoscopic, minimally invasive technique to the posterior surgical approach for cervical degenerative disease. The authors studied the microendoscopic posterior approach using the METRx system in both cadaver models and in clinical cases. This new technique needs only a small surgical route thus reducing damage to the paraspinous muscles. Moreover, this technique provides a clear view of the operating points, because of the oblique view angle of the endoscope. This technique is feasible for not only radiculopathy but also myelopathy caused by segmental canal stenosis. Posterior cervical decompression with this system is technically feasible and should be beneficial for reducing post-operative morbidity and spine deformity. This report deals with cases of cervical radiculopathy and segmental canal stenosis operated on with this system as well as the key points of this surgical procedure.

Inference of Marine Stratus Cloud Optical Depths from Satellite Measurements: Does 1D Theory Apply?

The validity of plane-parallel (1D) radiative transfer theory for cloudy atmospheres is examined by directly comparing calculated and observed visible reflectances for one month of Global Area Coverage Advanced Very High Resolution Radiometer satellite observations of marine stratus cloud layers off the coasts of California, Peru, and Angola. Marine stratus are an excellent testbed, as they arguably are the closest to plane-parallel found in nature. Optical depths in a 1D radiative transfer model are adjusted so that 1D model reflectances match those observed at nadir on a pixel-by-pixel basis. The 1D cloud optical depth distributions are then used in the plane-parallel model to generate reflectance distributions for different sun–earth–satellite viewing geometries. These reflectance distributions are directly compared with the observations. Separate analyses are performed for overcast and broken cloud layers as identified by the spatial coherence method. When 1D reflectances are directly compared with observations at different view angles, relative differences are generally small (≲10%) in the backscattering direction for solar zenith angles ≲60° and show no systematic view angle dependence. In contrast, 1D reflectances increase much more rapidly with view angle than the observed reflectances in the forward-scattering direction. Relative differences in the forward-scattering direction are ≈2–3 times larger than in the backscattering direction. At solar zenith angles ≳60°, the 1D model underestimates observed reflectances at nadir by 20%–30% and overestimates reflectances at the most oblique view angles in the forward scattering direction by 15%–20%. Consequently, when inferred on a pixel-by-pixel basis, nadir-derived cloud optical depths show a systematic increase with solar zenith angle, both for overcast and broken cloud layers, and cloud optical depths decrease with view angle in the forward scattering direction. Interestingly, in the case of broken marine stratocumulus, the common practice of assuming that pixels are overcast when they are not mitigates this bias to some extent, thereby confounding its detection. But even for broken clouds, the bias remains. Because of the nonlinear dependence of cloud albedo on cloud optical depth, errors in cloud optical depth lead to large errors in cloud albedo—and therefore energy budget calculations—regardless of whether cloud layers are overcast or broken. These findings suggest that as a minimum requirement, direct application of the plane-parallel model approximation should be restricted to moderate–high sun elevations and to view angles in the backscattering direction. Based on Monte Carlo simulations, the likely reason for the discrepancies between observed radiances and radiances calculated on the basis of 1D theory is because real clouds have inhomogeneous (i.e., bumpy) tops.

Angular dependence of observed reflectances: A comparison with plane parallel theory

In this study, a direct comparison between plane parallel model calculations and one year of Earth Radiation Budget Satellite (ERBS) shortwave scanner observations over ocean between 30°S and 30°N is performed. Considering only cloud‐contaminated pixels, plane parallel model calculations are first normalized to observations at nadir at high Sun elevations on a pixel‐by‐pixel basis by adjusting cloud fraction and cloud optical depth. These are then used to generate plane parallel model reflectances as a function of view, solar zenith, and relative azimuth angle, which are directly compared with the observations on a statistical basis. At moderate to high solar elevations (θo≲63°), the relative view angle dependence of plane parallel reflectances remains, on a statistical basis, within ≈10% (relative difference) of that of the observations. For larger solar zenith angles, however, observed and plane parallel reflectances show systematic differences at all view angles that increase with increasing solar zenith angle. Provided atmospheric effects above the cloud are accounted for in the calculations, observed reflectances exceed plane parallel values in the backscattering direction by roughly a constant amount at all view angles for θo≳63°. In the forward scattering direction, observed reflectances generally fall within the range of plane parallel model values (for a range of model assumptions) but show a very different view angle dependence; observed reflectances level off at view angles between 60° and 73°, whereas the calculations increase steadily. When stratified by pixel brightness, the plane parallel model generally provides a better approximation to the observed reflectance dependence on view angle for darker (i.e., optically thinner) pixels than for brighter (optically thicker) pixels. For the brightest pixels, reflectance differences are largest at nadir in the backscattering direction and at oblique view angles in the forward direction. Overall, the relative dependence on azimuthal angle is similar for the observations and plane parallel model, irrespective of cloud thickness. Neglecting pixel area expansion with view angle in the calculations is shown to have only a minor influence. Finally, the marked difference in the reflectance dependence on solar zenith angle between observations and calculations is suggested as the likely reason why the principle of directional reciprocity applied to satellite measurements breaks down at ERBS pixel scales.

The MISR Calibration Program

The Multiangle Imaging SpectroRadiometer (MISR) is currently under development for NASA's Earth Observing System. The instrument consists of nine pushbroom cameras, each with four spectral bands in the visible and near-infrared. The cameras point in different view directions to provide measurements from nadir to highly oblique view angles in the along-track plant. Multiple view-angle observations provide a unique resource for studies of clouds, aerosols, and the surface. MISR is built to challenging radiometric and geometric performance specifications. Radiometric accuracy, for example, must be within ±3%/1σ, and polarization insensitivity must be better than ±1%. An onboard calibrator (OBC) provides monthly updates to the instrument gain coefficients. Spectralon diffuse panels we used within the OBC to provide a uniform target for the cameras to view. The absolute radiometric scale is established both preflight and in orbit through the use of detector standards. During the mission, ground data processing to accomplish radiometric calibration, geometric rectification and registration of the nine view-angle imagery, and geophysical retrievals will proceed in an automated fashion. A global dataset is produced every 9 days. This paper details the preflight characterization of the MISR instrument, the design of the OBC, and the radiance product processing.

Observation and Recording of Laryngeal View by Use of a Modified Oblique-angled Type Telescope

A new system for observation and recording of the laryngeal view has been attained by use of a mcdified oblique-angled type telescope conbined with a powerful light source. The scope can be connected to an ordinary 35mm camera with a motor-drive device, a 16mm cine-camera, or a video-TV camera.The laryngeal view through the new scope is quite similar with that obtained by an indirect laryngoscopy. The oblique view angle is 70° for better visualization of the area near the anterior commissure of the vocal folds. We could obtain a clear picture of the laryngeal patholgy even in a 5 y-o child without any difficulty. For still photos a slide-type film (ASA 100) was used with no pushing and successful pictures were obtained in full-size.It was considerd that our tystem is usefull not only for clinical but also for educational purposes.