Automatic Feature Detection Research Articles

Learning to Encode Timing: Mechanisms of Plasticity in the Auditory Brainstem

Mechanisms of plasticity have traditionally been ascribed to higher-order sensory processing areas such as the cortex, whereas early sensory processing centers have been considered largely hard-wired. In agreement with this view, the auditory brainstem has been viewed as a nonplastic site, important for preserving temporal information and minimizing transmission delays. However, recent groundbreaking results from animal models and human studies have revealed remarkable evidence for cellular and behavioral mechanisms for learning and memory in the auditory brainstem.

Alternative electrode positioning for monitoring ECG with telemetry in freely moving rats

Implantable radio‐telemetry allows researchers to collect chronic electrocardiographic (ECG) data from freely moving animals in their home cage. ECG data is typically collected using wire electrodes placed subcutaneously in a lead II configuration. Unfortunately, subcutaneous leads in an ambulatory animal can sometimes be characterized by low signal amplitude, inconsistent waveform characteristics, and high background noise during voluntary muscle movement. These properties can contribute to the inability of researchers to consistently rely on automated software analysis packages to analyze ECG waveforms. This study compares five methods of ECG lead attachment: three using lead II electrode configuration, one previously published base/apex arrangement (Sgoifo, 1996) and a novel trans‐esophageal coaxial ECG lead. The ECG data were analyzed using automated feature detection software (P3P 4.9) and recommendations are presented for optimizing electrode placement for chronic telemetric monitoring of ECG in rats and mice.

Modification of fractal algorithm for oil spill detection from RADARSAT-1 SAR data

This paper introduces a modified formula for the fractal box counting dimension. The method is based on utilization of the probability distribution formula in the fractal box count. The purpose of this method is to use it for the discrimination of oil spill areas from the surrounding features, e.g., sea surface and look-alikes in RADARSAT-1 SAR Wide beam mode (W1) and Standard beam mode (S2) data have been collected under different wind speeds. The results show that the new formula of the fractal box counting dimension is able to discriminate between oil spills, look-alike areas and pixels of the size of a single ship. The W1 mode data illustrate an error standard deviation of 0.05, thus performing a better discrimination of oil spills as compared to S2 mode data. We conclude that automatic detection and discrimination of oil spill and other sea surface features can be opertionalized by using the new formula for fractal box counting.

Globally aligned photomosaic of the Lucky Strike hydrothermal vent field (Mid‐Atlantic Ridge, 37°18.5′N): Release of georeferenced data, mosaic construction, and viewing software

We present a georeferenced photomosaic of the Lucky Strike hydrothermal vent field (Mid‐Atlantic Ridge, 37°18′N). The photomosaic was generated from digital photographs acquired using the ARGO II seafloor imaging system during the 1996 LUSTRE cruise, which surveyed a ∼1 km2 zone and provided a coverage of ∼20% of the seafloor. The photomosaic has a pixel resolution of 15 mm and encloses the areas with known active hydrothermal venting. The final mosaic is generated after an optimization that includes the automatic detection of the same benthic features across different images (feature‐matching), followed by a global alignment of images based on the vehicle navigation. We also provide software to construct mosaics from large sets of images for which georeferencing information exists (location, attitude, and altitude per image), to visualize them, and to extract data. Georeferencing information can be provided by the raw navigation data (collected during the survey) or result from the optimization obtained from image matching. Mosaics based solely on navigation can be readily generated by any user but the optimization and global alignment of the mosaic requires a case‐by‐case approach for which no universally software is available. The Lucky Strike photomosaics (optimized and navigated‐only) are publicly available through the Marine Geoscience Data System (MGDS, http://www.marine‐geo.org). The mosaic‐generating and viewing software is available through the Computer Vision and Robotics Group Web page at the University of Girona (http://eia.udg.es/∼rafa/mosaicviewer.html).

Fuzzy logic techniques for blotch feature evaluation in dermoscopy images

Blotches, also called structureless areas, are critical in differentiating malignant melanoma from benign lesions in dermoscopy skin lesion images. In this paper, fuzzy logic techniques are investigated for the automatic detection of blotch features for malignant melanoma discrimination. Four fuzzy sets representative of blotch size and relative and absolute blotch colors are used to extract blotchy areas from a set of dermoscopy skin lesion images. Five previously reported blotch features are computed from the extracted blotches as well as four new features. Using a neural network classifier, malignant melanoma discrimination results are optimized over the range of possible alpha-cuts and compared with results using crisp blotch features. Features computed from blotches using the fuzzy logic techniques based on three plane relative color and blotch size yield the highest diagnostic accuracy of 81.2%.

Automated diagnosis of fetal alcohol syndrome using 3D facial image analysis

Use three-dimensional (3D) facial laser scanned images from children with fetal alcohol syndrome (FAS) and controls to develop an automated diagnosis technique that can reliably and accurately identify individuals prenatally exposed to alcohol. A detailed dysmorphology evaluation, history of prenatal alcohol exposure, and 3D facial laser scans were obtained from 149 individuals (86 FAS; 63 Control) recruited from two study sites (Cape Town, South Africa and Helsinki, Finland). Computer graphics, machine learning, and pattern recognition techniques were used to automatically identify a set of facial features that best discriminated individuals with FAS from controls in each sample. An automated feature detection and analysis technique was developed and applied to the two study populations. A unique set of facial regions and features were identified for each population that accurately discriminated FAS and control faces without any human intervention. Our results demonstrate that computer algorithms can be used to automatically detect facial features that can discriminate FAS and control faces.

Agrarian landscapes linear features detection from LiDAR: application to artificial drainage networks

Linear features of agrarian landscapes are the anthropogenic elements such as hedgerows, ditches, and bench terraces that strongly impact agrarian areas' environmental behaviour, especially the ecological and hydrological areas. The need to map these linear elements for environmental impact assessments of agrarian areas is thus increasing since these maps limit the developments of spatial indicators and spatially distributed modelling. Until now, no generic remote sensing methodology has been proposed for such mapping purposes. This research was designed to assess the use of airborne LiDAR data for agrarian landscape linear features mapping. We proposed a methodology that uses LiDAR data in three steps. We first estimated elevation profiles from LiDAR points on a set of pre‐located sites. We secondly performed profile shape description with wavelet transform or a watershed algorithm. Finally, we classified the profiles using classification trees with predictors coming from shape analysis. Methodology accuracies were calculated for a ditch network detection problem in a Mediterranean vineyard landscape. LiDAR Toposys data and field survey data for ditch location were collected in June 2002. As ditches are always located on field boundary lattices, elevation profiles were only computed on field boundary sites. Methodologies, using wavelets or the watershed algorithm, gave similar accuracies. Overall accuracy is about 70% with a high ditch omission rate (50%) but low commission rate (15%). The omissions conform to those obtained when performing visual classification of profiles. This high omission rate in ditch detection is therefore due to LiDAR data, not methods. Dense vegetation over ditches during the summertime and the specific LiDAR points spatial sampling design explain these omissions. However, the proposed methodology, especially using wavelets transform, looks transposable for the automatic detection or characterization of other agrarian linear features.

Development of a quantitative, validated Capillary electrophoresis‐time of flight – mass spectrometry method with integrated high‐confidence analyte identification for metabolomics

A CE-MS method was developed and validated for the quantitative analysis of negatively charged metabolites by making use of the high mass accuracy and the quantitation capabilities of a TOF mass analyzer in combination with automated feature extraction and database search. Metabolites of the central carbon metabolism were quantified with an LOD and lower LOQ (LLOQ) of 0.2-2 and 1-4 microM, respectively. The method was used to elucidate metabolic changes in the Escherichia coli deletion mutant PntAB-UdhA that lacks nicotinamide nucleotide transhydrogenase function, under both stationary and exponential growth conditions. The reproducibility of metabolite extraction and CE-TOF-MS analysis ranged from 3.7 to 22.7 and 7.9 to 22.6%, respectively, while the biological variance was 3.4-31.3%. We observed significant differences in metabolite abundance, particularly in the citrate cycle, between wild-type and mutant E. coli. Overall, more than 600 features were found by automated feature detection, which resulted in approximately 150 high-confidence metabolite identifications. Concomitant analyses with two different GC-MS methods allowed not only crossvalidation of the quantitative results obtained by the various methods, but also led to a more comprehensive coverage of the E. coli metabolome.

Automatic detection of anisotropic features on rock surfaces

Surface roughness is an important rock property that is measured for structural geology and engineering purposes. We have developed an automatic technique to detect anisotropic features on rock faces based on fractal analysis. The analysis method has been applied to synthetic surfaces, and to digitally mapped point clouds of natural rock surfaces shaped by weathering, fault wear, and mining. We illustrate the technique using field examples from Permian sandstones containing brittle shear zones in northeast England, the surface of a neotectonic fault in Turkey, Proterozoic quartzite from central Australia, and Devonian Quartzite in an aggregate quarry in Germany. Roughness analysis of these natural examples suggests that a significant change of roughness value, anisotropy, and anisotropy direction can exist across scale. Our analysis method represents a step toward developing a toolkit to automatically detect and interpret surface characteristics from digitally acquired data sets. It has widespread potential for applications in rock engineering and the geosciences.

ESVM: Evolutionary support vector machine for automatic feature selection and classification of microarray data

An optimal design of support vector machine (SVM)-based classifiers for prediction aims to optimize the combination of feature selection, parameter setting of SVM, and cross-validation methods. However, SVMs do not offer the mechanism of automatic internal relevant feature detection. The appropriate setting of their control parameters is often treated as another independent problem. This paper proposes an evolutionary approach to designing an SVM-based classifier (named ESVM) by simultaneous optimization of automatic feature selection and parameter tuning using an intelligent genetic algorithm, combined with k-fold cross-validation regarded as an estimator of generalization ability. To illustrate and evaluate the efficiency of ESVM, a typical application to microarray classification using 11 multi-class datasets is adopted. By considering model uncertainty, a frequency-based technique by voting on multiple sets of potentially informative features is used to identify the most effective subset of genes. It is shown that ESVM can obtain a high accuracy of 96.88% with a small number 10.0 of selected genes using 10-fold cross-validation for the 11 datasets averagely. The merits of ESVM are three-fold: (1) automatic feature selection and parameter setting embedded into ESVM can advance prediction abilities, compared to traditional SVMs; (2) ESVM can serve not only as an accurate classifier but also as an adaptive feature extractor; (3) ESVM is developed as an efficient tool so that various SVMs can be used conveniently as the core of ESVM for bioinformatics problems.

Non-Linear Filtering and Limiting in High Order Methods for Ideal and Non-Ideal MHD

The adaptive nonlinear filtering and limiting in spatially high order schemes (Yee et al. J. Comput. Phys. 150, 199–238, (1999), Sjögreen and Yee, J. Scient. Comput. 20, 211–255, (2004)) for the compressible Euler and Navier–Stokes equations have been recently extended to the ideal and non-ideal magnetohydrodynamics (MHD) equations, (Sjögreen and Yee, (2003), Proceedings of the 16th AIAA/CFD conference, June 23–26, Orlando F1; Yee and Sjögreen (2003), Proceedings of the International Conference on High Performance Scientific Computing, March, 10–14, Honai, Vietnam; Yee and Sjögreen (2003), RIACS Technical Report TR03. 10, July, NASA Ames Research Center; Yee and Sjögreen (2004), Proceedings of the ICCF03, July 12–16, Toronto, Canada). The numerical dissipation control in these adaptive filter schemes consists of automatic detection of different flow features as distinct sensors to signal the appropriate type and amount of numerical dissipation/filter where needed and leave the rest of the region free from numerical dissipation contamination. The numerical dissipation considered consists of high order linear dissipation for the suppression of high frequency oscillation and the nonlinear dissipative portion of high-resolution shock-capturing methods for discontinuity capturing. The applicable nonlinear dissipative portion of high-resolution shock-capturing methods is very general. The objective of this paper is to investigate the performance of three commonly used types of discontinuity capturing nonlinear numerical dissipation for both the ideal and non-ideal MHD.

Production of Filaments by Surges

We have recently developed a new method for automatic detection of solar surges and other dark dynamic features. For the first time we find that some filaments can be quickly formed by trapping the cold material supplied by surges originating from the chromosphere. Two clear examples are presented in this Letter, showing the special process for new filament formation by surge injection. The data used here were taken from Hida Observatory and the Global Hα Network operated by Big Bear Solar Observatory. Both the filaments existed on the solar surface for not less than 20 hr, with an average length about 200'', and had obvious helical structures and barbs. The surge material was injected from one terminal along the main axis of the filaments or the filament channels. We conclude that there are two necessary conditions for new filament formation by surges at one location: (1) an "empty" filament channel, or magnetic trap, and (2) enough mass supplied by surge activity. In most other surge-filament events, apart from our two examples, the surges are observed moving toward and acting on a preexisting nearby filament. The close relationship between surge activity and filament formation and maintenance suggests that there should be a direct link between the filament axial fields and the large-scale background fields along which the surge material can be driven into the filament channel. On the other hand, it also supports the idea that the frequent injection of flow from below is an important way to convey mass and energy into the corona through magnetic reconnections driven by successive emerging flux and converging flow.

Shadow Analysis in High-Resolution Satellite Imagery of Urban Areas

High-resolution satellite imagery (HRSI) offers great possibilities for urban mapping. Unfortunately, shadows cast by buildings in high-density urban environments obscure much of the information in the image leading to potentially corrupted classification results or blunders in interpretation. Although significant research has been carried out on the subject of shadowing in remote sensing, very few studies have focused on the particular problems associated with high-resolution satellite imaging of urban areas. This paper reviews past and current research and proposes a solution to the problem of automatic detection and removal of shadow features. Tests show that although detection and removal of shadow features can lead to improved image quality, results can be image-dependent.

Automatic detection and verification of solar features

AbstractA fast hybrid system for the automated detection and verification of active regions (plages) and filaments in solar images is presented in this paper. The system combines automated image processing with machine learning. The imaging part consists of five major stages. The solar disk is detected in the first stage, using a morphological hit‐miss transform, watershed transform and Filling algorithm. An image‐enhancement technique is introduced to remove the limb‐darkening effect and intensity filtering is implemented followed by a modified region‐growing technique to detect the regions of interest (RoI). The algorithms are tested on H‐α and CA II K3‐line solar images that are obtained from Meudon Observatory, covering the period from July 2, 2001 till August 4, 2001. The detection algorithm is fast and it achieves false acceptance rate (FAR) error rate of 67% and false rejection rate (FRR) error rate of 3% for active regions, and FAR error rate of 19% and FRR error rate of 14% for filaments, when compared with the manually detected filaments in the synoptic maps. The detection performance is enhanced further using a neural network (NN), which is trained on statistical features extracted from the RoI and non‐RoI. With the use of this combination the FAR has dropped to 2% for active regions and 4% for filaments.© 2006 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 15, 199–210, 2005

Multi-spectral front maps for automatic detection of ocean colour features from SeaWiFS

The aim of this research is to automatically detect and visualize dynamic ocean colour phenomena such as algae blooms, fronts and eddies from a sequence of cloudy satellite images. The composite front map methodology has been extended to combine feature observations from multiple ocean colour and temperature products in a single map, to explore the interaction between physical and biological oceanic processes. Sample maps showing chlorophyll, sediment and sea surface temperature fronts are presented, derived from long sequences of cloud-affected Advanced Very High Resolution Radiometer (AVHRR) and Sea viewing Wide Field of view Sensor (SeaWiFS) data. The successful detection and animation of many ocean colour features suggests a valuable application to summarize the increased multi-spectral data provided by the National Aeronautics & Space Administration's (NASA) Moderate Resolution Imaging Spectroradiometer (MODIS) instrument.

Visible and infrared spectroscopy of minerals and mixtures with the OMEGA/MARS-EXPRESS instrument

We present the reflectance measurements made with OMEGA during ground calibration on a selection of minerals and mixtures considered as candidates for constituents of the Mars surface. OMEGA is the visible and infrared mapping spectrometer designed to fly on-board the ESA/MARS-EXPRESS probe. These measurements allow to evaluate the instrument performances in identifying minerals as major or minor constituents (such as carbonates and sulfates). We use a new algorithm for the automatic detection of spectral features, and a multiple scattering model, to derive the mineralogical composition, the relative abundances and the mean grain size from the sample reflectance spectra. We discuss the implication on the OMEGA capability to meet its scientific objectives. In particular, the present work emphasize the interest of the 3– 5 μm spectral range as a diagnostic one in the regions of Mars where the thermal emission will be low enough.

Geometric Snakes for Triangular Meshes

Feature detection is important in various mesh processing techniques, such as mesh editing, mesh morphing, mesh compression, and mesh signal processing. In spite of much research in computer vision, automatic feature detection even for images still remains a difficult problem. To avoid this difficulty, semi-automatic or interactive techniques for image feature detection have been investigated. In this paper, we propose a geometric snake as an interactive tool for feature detection on a 3D triangular mesh. A geometric snake is an extension of an image snake, which is an active contour model that slithers from its initial position specified by the user to a nearby feature while minimizing an energy functional. To constrain the movement of a geometric snake onto the surface of a mesh, we use the parameterization of the surrounding region of a geometric snake. Although the definition of a feature may vary among applications, we use the normal changes of faces to detect features on a mesh. Experimental results demonstrate that geometric snakes can successfully capture nearby features from user-specified initial positions.

Use of a simple pattern recognition approach for the detection of ridge lines and stripes

AbstractPatterns often sought in meteorological fields are ridge lines or narrow zones of relatively high or low values (e.g. stripes or bands in satellite imagery). A method for the automatic detection of such features is presented, using input fields given on regular grids. The fundamental idea is that a grid point is a plausible position for a significant line where not more than two (out of eight) neighbours exhibit better suitability to lie on that very line. Locations with three or more superior neighbours are considered unsuitable, or at least less qualified, for representing positions of the line. This criterion derives from the concept that a significant line passing through a point normally enters the latter from one of the adjacent points and leaves through another neighbour, with these two special neighbours permissibly having superior values of the appropriate test variable. It is empirically demonstrated for selected applications within the field of air mass analysis (including objective front positioning on the basis of numerical model output) that the identification of points satisfying the formulated ‘2‐out‐of‐8’ criterion is, when linked with simple post‐processing steps, sufficient to accomplish a good recognition of the sought patterns. Copyright © 2002 Royal Meteorological Society.

Detection and modeling of buildings from multiple aerial images

Automatic detection and description of cultural features, such as buildings, from aerial images is becoming increasingly important for a number of applications. This task also offers an excellent domain for studying the general problems of scene segmentation, 3D inference, and shape description under highly challenging conditions. We describe a system that detects and constructs 3D models for rectilinear buildings with either flat or symmetric gable roofs from multiple aerial images; the multiple images, however, need not be stereo pairs (i.e., they may be acquired at different times). Hypotheses for rectangular roof components are generated by grouping lines in the images hierarchically; the hypotheses are verified by searching for presence of predicted walls and shadows. The hypothesis generation process combines the tasks of hierarchical grouping with matching at successive stages. Overlap and containment relations between 3D structures are analyzed to resolve conflicts. This system has been tested on a large number of real examples with good results, some of which are included in the paper along with their evaluations.

Automatic identification of time series features for rule-based forecasting

Rule-based forecasting (RBF) is an expert system that uses features of time series to select and weight extrapolation techniques. Thus, it is dependent upon the identification of features of the time series. Judgmental coding of these features is expensive and the reliability of the ratings is modest. We developed and automated heuristics to detect six features that had previously been judgmentally identified in RBF: outliers, level shifts, change in basic trend, unstable recent trend, unusual last observation, and functional form. These heuristics rely on simple statistics such as first differences and regression estimates. In general, there was agreement between automated and judgmental codings for all features other than functional form. Heuristic coding was more sensitive than judgment and consequently, identified more series with a certain feature than judgmental coding. We compared forecast accuracy using automated codings with that using judgmental codings across 122 series. Forecasts were produced for six horizons, resulting in a total of 732 forecasts. Accuracy for 30% of the 122 annual time series was similar to that reported for RBF. For the remaining series, there were as many that did better with automated feature detection as there were that did worse. In other words, the use of automated feature detection heuristics reduced the costs of using RBF without negatively affecting forecast accuracy.