Rail Corrugation Detection and Characterization Using Computer Vision

Rail corrugation appears as oscillatory wear on the rail surface caused by the interaction between the train wheels and the railway. Corrugation shortens railway service life and forces early rail replacement. Consequently, service can be suspended for days during rail replacement, adversely affecting an important means of transportation. We propose an inspection method for rail corrugation using computer vision through an algorithm based on feature descriptors to automatically distinguish corrugated from normal surfaces. We extract seven features and concatenate them to form a feature vector obtained from a railway image. The feature vector is then used to build support vector machine. Data were collected from seven different tracks as video streams acquired at 30 fps. The trained support vector machine was used to predict test frames of rails as being either corrugated or normal. The proposed method achieved a high performance, with 97.11% accuracy, 95.52% precision, and 97.97% recall. Experimental results show that our method is more effective in identifying corrugated images than reference state-of the art works.

ℤ3-actions on Horikawa surfaces

Minimal algebraic surfaces of general type [Formula: see text] such that [Formula: see text] are called Horikawa surfaces. In this note, [Formula: see text]-actions on Horikawa surfaces are studied. The main result states that given an admissible pair [Formula: see text] such that [Formula: see text], all the connected components of Gieseker’s moduli space [Formula: see text] contain surfaces admitting a [Formula: see text]-action. On the other hand, the examples considered allow us to produce normal stable surfaces that do not admit a [Formula: see text]-Gorenstein smoothing. This is illustrated by constructing non-smoothable normal surfaces in the KSBA-compactification [Formula: see text] of Gieseker’s moduli space [Formula: see text] for every admissible pair [Formula: see text] such that [Formula: see text]. Furthermore, the surfaces constructed belong to connected components of [Formula: see text] without canonical models.

Observers Agreement in Perception of Non-Cavitated Approximal Dental Caries by Intraoral Digital CCD Radiography at Different Exposure Parameters and Corresponding Required Radiation Dose

Objectives: To evaluate the diagnostic accuracy of Charged Coupled device (CCD) in detection of Non- Cavitated Approximal caries at different exposure parameters in relation to radiation dose in vitro. Study Design: Seventy-eight surfaces of extracted teeth were inserted in acid gel to create non-cavitated proximal caries with different depth, and then Radiographs have been taken to all teeth by CCD sensor. Radiographs were interpreted by three observers. The lesions were classified as (N) No lesion, (D1) Less than ½ enamel thickness, (D2) more than halfway of enamel but not involve DEJ. (D3) Dentin caries. Teeth were randomly selected for histological analysis after consensus from three oral and maxillofacial radiologists as Gold standard. The corresponding radiation dose was measured by unfors meter device at different exposure parameters. Results: The histological examination showed that the distribution of lesions was 39.8% Sound, both enamel lesions are equal 17.8%, Dentin lesions 24.6. The sensitivity and specificity of CCD to detect normal surfaces were 0.95, D1 was 0.37, D2 was 0.74 and D3 was 0.86. As the lesions depth increased, the sensitivity increased. The higher image quality was produced by using exposure parameters (70 KvP, 160 ms) and (70 KvP, 200 ms). While, (60 KvP, 200 ms) and (60 KvP, 250 ms) produced the worse image quality. Conclusion: Regard the balance between the higher diagnostic accuracy of digital images and minimum radiation dose: using exposure parameters as (70 KvP, 160 ms) is considered the best image quality and relative dose (81 mSv). While, (70 KvP, 125ms) and (66 KvP, 160 ms) are little bit lower quality and corresponding dose are (63), (73) respectively. Although (70 KvP, 200 ET) produce higher image quality but its relative dose is high (101mSv).

Hyperspectral, Time-Resolved, Fluorescence Imaging System for Large Sample Sizes: Part I. Development of a High-Energy Line-Illumination Source

To reduce the risk of foodborne illnesses, produce fields are visually surveyed prior to harvest for signs of fecal contamination. To improve the efficacy of surveys, a hyperspectral, line-scan, laser-induced fluorescence imaging system was developed. The goal is to incorporate the imaging system into a field-deployable apparatus to survey produce fields. The system includes a gated intensified camera, a spectral adapter, a 355 nm pulsed laser, and a Powell lens that is used to expand the laser beam into a line-illumination source. Software was developed to facilitate alignment of the Powell lens with the laser beam and the resulting line-illumination profile with the line-imaging field. Comparisons were made of uniformity and efficiency measures for regions within illumination profiles that corresponded to the camera field of view for the Powell lens and previously developed simple and homogenizing expansion optics. Spatial and temporal uniformity measures were similar for the Powell and homogenizing optics, and both were better than for simple optics; however, total efficiency was better for Powell compared to homogenizing optics at 28.5% and 3.0%, respectively. After spectral calibrations, theoretical and measured spectral peaks of five fluorescent standards were identical. Images of apples and spinach artificially contaminated with dilutions of dairy manure demonstrated high contrast. By selecting appropriate gate timing parameters, it was possible to create images in which responses from contamination sites were still evident but responses from normal surfaces were effectively extinguished. These results demonstrate that the system has potential to be used to detect sites of fecal contamination in produce fields. Keywords: Fecal detection, Fluorescence imaging, Food safety, Machine vision, Powell lens.