Minimum Mahalanobis Distance Research Articles

Evaluation of LISS-III Sensor Data of IRS-P6 Satellite for Detection Saline Soils (Case Study: Najmabad Region)

Soil Salinity has been a large problem in arid and semi arid regions. Preparation of such maps is useful for Natural resource managers. Old methods of preparing such maps require a lot of time and cost. Multi-spectral remotely sensed dates due to the broad vision and repeating of these imageries is suitable for provide saline soil maps. This investigation is conducted to provide saline soil maps with sensor LISS-III of IRS-P6 satellite data, in Najmabad of Savojbolagh. Satellite images belonging to 25 June 2006. For enhancement of images, salt Indices, Digital Elevation Model (DEM), False Color Composite imageries (FCC) and Principal Component Analysis (PCA), were used. Supervised classification method includes Box classifier, Minimum Distance, Minimum Mahalanobis Distance and Maximum Likelihood classifier, DEM, PCA1, PCA4 and Saline Indices (SI) were used. After classification, the class map salinity S0, S1, S2, S3 S4, were prepared. The results shows highest overall accuracy and kappa coefficient for the maximum Likelihood classifier estimate, respectively 99% and 97% and the lowest overall accuracy and kappa coefficient for PCA1 estimate, respectively 1% and 0% were obtained. Using Digital Elevation Model (DEM) also due to the difference in height position to the separation of saline lands is usefully. Most spectral interference relatedto non-saline soils and low saline soil. From among indices INT2 and PVI greatest ability to segregate is salty soils(especially classes S0 and S1).

Proposing a methodology in preparation of olive orchards map as an important medicinal plant in Iran by remote sensing (RS) and geographical information system (GIS)

Present study focuses on identification and mapping of olive in the part of Roodbar region, Guilan, Iran by IRS images and GIS. Two methods were evaluated to separate olive orchards spectrum reflex from the other surface covers. At the first method, upper and lower limit of digital number of olive orchards were determined by the addition and subtraction of standard deviation from the mean in each band and initial classified map of olive orchards was prepared in every band. The final map of olive orchards was prepared from crossing three initial maps of olive orchards. At the second method, olive orchards map was prepared by four models include: Box classifier, maximum likelihood, minimum distance and minimum Mahalanobis distance. Methods accuracy was evaluated from crossing the map of training points (pixel) with olive orchards map. The results indicated that in classification of less-condensed olive orchards, because of spectrum wave interference of olive green canopy and the soil zone between the canopy cover, the interference of digital number of low-condensed olive is observed with the other vegetation cover and bare lands. There was this issue even for wave interference of low-condensed olive with urban and residential regions. There is the interference of spectrum reflexes between the agriculture land and olive. This issue also exists for different methods of supervised classification. Key words: Spectrum reflex, olive, training points, map.

Providing poplar plantation map by Indian remote sensing (IRS) satellite imagery in Northern Iran

Using satellite imagery processing technique, we can separate planting poplar lands with a low cost, high speed and precision. The present study was conducted in northern Iran using Indian Remote Sensing (IRS) satellite data in July 2006 and ILWIS software to identify the cultivated surfaces of poplar. Field visitation recorded 548 ground control points by GPS in the way of features in various areas of Guilan Province. Ground control point map was overlaid on satellite color composite (sample sat) as training points map (pixels). In supervised classification, box classifier, maximum likelihood, minimum distance and minimum mahalanobis distance were used to determine user’s accuracy, overall accuracy and kappa coefficient of each method, separately. Results indicate that spectrum reflex of poplar species is different from other vegetations and separable; but it has very close interference with conifer forest, natural forest, rice field and canebrake. By determining suitable training points, poplar cultivated surfaces could be identified. Among the classification methods, maximum likelihood with 91.48% of overall accuracy and 90.75% of kappa coefficient is the best method for separating planting poplar lands compared to the other methods. Key words: Poplar planting, supervised classification, maximum likelihood method, Indian remote sensing (IRS), Guilan.

Comparison of hyperspectral imagery with aerial photography and multispectral imagery for mapping broom snakeweed

Broom snakeweed (Gutierrezia sarothrae (Pursh) Britt. & Rusby) is one of the most widespread and abundant rangeland weeds in western North America. The objectives of this study were to evaluate airborne hyperspectral imagery and compare it with aerial colour-infrared (CIR) photography and multispectral digital imagery for mapping broom snakeweed infestations. Airborne hyperspectral imagery along with aerial CIR photographs and digital CIR images was acquired from a rangeland area in south Texas. The hyperspectral imagery was transformed using minimum noise fraction (MNF) and then classified using minimum distance, Mahalanobis distance, maximum likelihood, and spectral angle mapper (SAM) classifiers. The digitized aerial photographs and the digital images were respectively mosaicked as one photographic image and one digital image; these were then classified using the same classifiers. Accuracy assessment showed that the maximum likelihood classifier performed the best for the three types of images. The best overall accuracies for three-class classification maps (snakeweed, mixed woody and mixed herbaceous) were 91.0%, 92.5%, and 95.0%, respectively, for the CIR photographic image, the digital CIR image and the MNF-transformed hyperspectral image. Kappa analysis showed that there were no significant differences in maximum likelihood-based classifications among the three types of images. These results indicate that airborne hyperspectral imagery along with aerial photography and multispectral imagery can be used for monitoring and mapping broom snakeweed infestations on rangelands.

Proposed methods of image recognition depend on the PCA

This paper suggest two method of recognition, these methods depend on the extraction of the feature of the principle component analysis when applied on the wavelet domain(multi-wavelet). First method, an idea of increasing the space of recognition, through calculating the eigenstructure of the diagonal sub-image details at five depths of wavelet transform is introduced. The effective eigen range selected here represent the base for image recognition. In second method, an idea of obtaining invariant wavelet space at all projections is presented. A new recursive from that represents invariant space of representing any image resolutions obtained from wavelet transform is adopted. In this way, all the major problems that effect the image and change its characteristic are solved through calculating invariant eigen range of the recursive resolution forms of all sub-images coefficient. These approaches employed here as multi-wavelet transform identifier with minimum Mahalanobis distance. All method recognition proposed in this paper are applied on different images. Different tables of image recognition resulted in accurate and fast.

Applying image transformation and classification techniques to airborne hyperspectral imagery for mapping Ashe juniper infestations

Ashe juniper (Juniperus ashei Buchholz) in excessive coverage reduces forage production, interferes with livestock management, and degrades watersheds and wildlife habitat on infested rangelands. The objective of this study was to apply minimum noise fraction (MNF) transformation and different classification techniques to airborne hyperspectral imagery for mapping Ashe juniper infestations. Hyperspectral imagery with 98 usable bands covering a spectral range of 475–845 nm was acquired from two Ashe juniper infested sites in central Texas. MNF transformation was applied to the hyperspectral imagery and the transformed imagery with the first 10 and 20 MNF bands was classified using four hard classifiers: minimum distance, Mahalanobis distance, maximum likelihood and spectral angle mapper (SAM). For comparison, the 10‐ and 20‐band MNF imagery was inversely transformed to noise‐reduced 98‐band imagery in the original data space, which was also classified using the four classifiers. Accuracy assessment showed that the first 10 MNF bands were sufficient for distinguishing Ashe juniper from associated plant species (mixed woody species and mixed herbaceous species) and other cover types (bare soil and water). Although the 20‐band MNF imagery provided better results for some classifications, the increase in overall accuracy was not statistically significant. Overall accuracy on the 10‐band MNF imagery varied from 88% for SAM to 93% for minimum distance for site 1 and from 84% for SAM to 94% for maximum likelihood for site 2. The 98‐band imagery derived from the 10‐band MNF imagery resulted in overall accuracy ranging from 91% for both SAM and Mahalanobis distance to 97% for maximum likelihood for site 1 and from 87% for SAM to 93% for minimum distance for site 2. Although both approaches produced comparable classification results, the MNF imagery required smaller storage space and less computing time. These results indicate that airborne hyperspectral imagery incorporated with image transformation and classification techniques can be a useful tool for mapping Ashe juniper infestations.

Sparse multinomial kernel discriminant analysis (sMKDA)

Dimensionality reduction via canonical variate analysis (CVA) is important for pattern recognition and has been extended variously to permit more flexibility, e.g. by “kernelizing” the formulation. This can lead to over-fitting, usually ameliorated by regularization. Here, a method for sparse, multinomial kernel discriminant analysis (sMKDA) is proposed, using a sparse basis to control complexity. It is based on the connection between CVA and least-squares, and uses forward selection via orthogonal least-squares to approximate a basis, generalizing a similar approach for binomial problems. Classification can be performed directly via minimum Mahalanobis distance in the canonical variates. sMKDA achieves state-of-the-art performance in terms of accuracy and sparseness on 11 benchmark datasets.

Evaluating airborne hyperspectral imagery for mapping waterhyacinth infestations

Waterhyacinth [Eichhornia crassipes (Mart.) Solms] is an exotic aquatic weed that often invades and clogs waterways in many tropical and subtropical regions of the world. The objective of this study was to evaluate airborne hyperspectral imagery and different image classification techniques for mapping waterhyacinth infestations on Lake Corpus Christi in south Texas. Hyperspectral imagery with bands in the visible to near-infrared region of the spectrum was acquired from two study sites and minimum noise fraction (MNF) transformation was used to reduce the spectral dimensionality of the imagery. Four classification methods, including minimum distance, Mahalanobis distance, maximum likelihood, and spectral angle mapper (SAM), were applied to the MNF-transformed imagery for distinguishing waterhyacinth from associated plant species (waterlettuce, mixed herbaceous species, and mixed woody species) and other cover types (bare soil and water). Accuracy assessment showed that overall accuracy varied from 79% for SAM to 96% for maximum likelihood for site 1 and from 84% for minimum distance to 95% for maximum likelihood for site 2. Kappa analysis showed that maximum likelihood was significantly better than the other three methods and that there were no significant differences in overall classifications among the other three methods. Producer's and user's accuracies for waterhyacinth based on maximum likelihood were 94% and 100%, respectively, for site 1 and 100% and 95% for site 2. These results indicate that airborne hyperspectral imagery incorporated with image transformation and classification techniques can be a useful tool for mapping waterhyacinth infestations.

Preictal state identification by synchronization changes in long-term intracranial EEG recordings

Objective There is accumulated evidence that mesial temporal lobe seizures are preceded by a preictal transition that evolves over minutes to hours. In the present study, we investigated these possible preictal changes in long-term intracranial recordings of five patients by a measure of phase synchronization. In order to clearly distinguish preictal changes from all the other interictal states, we developed an automatic extraction of representative patterns of interictal synchronization activity. This reference library was used to classify the successive synchronization patterns of long-term recordings into groups of similar patterns. Altered states of brain synchronization were identified as deviating from patterns in the reference library and were evaluated relative to the times of seizure onset in terms of sensitivity and specificity. Methods A phase-locking measure was estimated using a sliding window analysis on 15 frequency bands (2 Hz steps between 0 and 30 Hz), for all pairs of EEG channels in the epileptogenic temporal lobe (14–20 channels), over the entire data sets (total analyzed duration 305 h). The preictal identification encompasses three basic stages: (1) a preprocessing stage involving the determination of a reference library of characteristic interictal synchronization patterns using a K-means algorithm, and the identification of discriminant variables differentiating interictal from preictal states, (2) a classification stage of the synchronization pattern via a minimum Mahalanobis distance to the reference patterns, as well as detection of outliers, (3) an evaluation stage of the sensitivity and specificity of the detection by receiver-operating characteristic curves. Results In most of the cases (36 of 52 seizures, i.e. 70%), a specific state of brain synchronization can be observed several hours before the actual seizure. The changes involved both increases and decreases of the synchronization levels, occurring mostly within the 4–15 Hz frequency band, and were often localized near the primary epileptogenic zone. Conclusions The analysis of phase synchronization offers a way to distinguish between a preictal state and normal interictal activity. These findings suggest that brain synchronizations are preictally altered in the epileptogenic temporal lobe, inducing a pathological state of higher susceptibility for seizure activity. Significance Phase synchronization is capable of extracting information from the EEG that allow the definition of a preictal state. Although the proposed analysis does not constitute genuine seizure anticipation, these changes in neuronal synchronization may provide helpful information for prospective seizure warning.

Feature extraction and supervised classification of MR images to support proton radiation therapy of eye tumors

Proton therapy has the potential for high-precision radiotherapy of retinal tumors. However, the standardized eye models currently used do not fully account for the patient's individual anatomy. To better exploit the data provided by MR images, a model-based approach was used based on a database of eye models. A face recognition algorithm was advanced to define similarity criteria between the reference image and the actual image. After building a high-dimensional feature vector and using a training data set, the reference model was selected by using the minimum Mahalanobis distance between the image to be classified and the reference images.

NONLINEAR SIGNAL CLASSIFICATION

In this contribution, we show that the incorporation of nonlinear dynamical measures into a multivariate discrimination provides a signal classification system that is robust to additive noise. The signal library was composed of nine groups of signals. Four groups were generated computationally from deterministic systems (van der Pol, Lorenz, Rössler and Hénon). Four groups were generated computationally from different stochastic systems. The ninth group contained inter-decay interval sequences from radioactive cobalt. Two classification criteria (minimum Mahalanobis distance and maximum Bayesian likelihood) were tested. In the absence of additive noise, no errors occurred in a within-library classification. Normally distributed random numbers were added to produce signal to noise ratios of 10, 5 and 0 dB. When the minimum Mahalanobis distance was used as the classification criterion, the corresponding error rates were 2.2%, 4.4% and 20% (Expected Error Rate = 89%). When Bayesian maximum likelihood was the criterion, the error rates were 1.1%, 4.4% and 21% respectively. Using nonlinear measures an effective discrimination can be achieved in cases where spectral measures are known to fail. Most classification errors occurred at low signal to noise ratios when a stochastic signal was misclassified into a different group of stochastic signals. When the within-library classification exercise is limited to the four groups of deterministic signals, no classification errors occurred with clean data, at SNR = 10 dB, or at SNR = 5 dB. A single classification error (Observed Error Rate = 2.5%, Expected Error Rate = 75%) occurred with both classification criteria at SNR = 0 dB.

Damage Identification by Discriminant Analysis Using Mahalanobis Distance.

The present paper proposes a new diagnostic tool for the structural health monitoring that employs a Mahalanobis Distance adopted in discriminant analysis. Structural health monitoring is a noticeable technology for aged civil structures. Most of the structural health monitoring systems adopts parametric method based on modeling or non-parametric method such as artificial neural networks or response surfaces. The conventional methods require FEM modeling of structure or a regression model. This modeling needs judgment of human, and it requires much costs. The present method does not require the process of modeling, in order to identify the damage level using the discriminant analysis. This suggest us, this technique is applicable to the health monitoring system, which identifies the damage of the structure, easily. In the present study, we developed the damage identification method using minimum Mahalanobis Distance. We applied this method to identifications of delamination crack of CFRP Plate. As a result, it was shown that this method is effective for identification of damages.

K-2330 統計的手法を用いたCFRP眉間はく離診断手法の比較検討(J03-6 ヘルスモニタリング)(J03 知的材料・構造システム)

Delamination cracks are invisible and cause decrease of compression strength of laminated composites. Therefore, health-monitoring system is required for laminates. The present study adopts an electric potential method for health monitoring of graphite/epoxy laminated composites. The electric potential method does not cause strength reduction and can be applied existing structures by low cost. The present study focuses on the applicability of the electric potential method for identification of delamination crack. In the present study, we developed the damage identification method using minimum Mahalanobis Distance as the method for identifying delamination from data. We applied this method to identifications of delamination crack of CFRP Plate and compared this method with Response surface method. As a result, it was shown that this method is effective for identification of damages.

Computer-based discrimination of abnormal stomachs using radiograms.

The possibility of automating mass screening of gastric disease was studied by applying statistical pattern recognition methods to features extracted from standing position-anteroposterior radiograms by two approaches. A region of interest was defined to encompass the deformity of the apex or the sulcus angularis. The ability of the features to classify normal, ulcerated, and cancerous cases was evaluated by a typical method of discriminant analysis, the minimum Mahalanobis distance classification. The error classification rate was estimated by the leaving-one-out method. Best feature sets were selected by the forward sequential search. The separability of the samples and the properties of the features were inspected also by cluster analysis. For the selected feature sets, the error classification rate obtained by the discriminant analysis agreed well with results of the cluster analysis and of inspection by physicians. The selected features tended to belong to different feature clusters. This seems to validate the feature selection.

Values in Parochial and Public Schools: Alike or Different?

Rokeach's Value Survey Scale was administered to seventh and eighth grade students attending Catholic ( n — 61), Episcopalian ( n — 113), Friends ( n = 29), and Public ( n — 40) Schools. The four-group stepwise discriminant function carried out found that seven value rankings maximized the minimum Mahalanobis distance between group pairs: self-controlled, salvation, social recognition, national security, courageous, equality, and obedience. Two statistically significant functions were derived, Concern for Eternal Life and Recognition Reliability and Acceptance. The scale correctly classified 56% of the students in all groups with 70.5% of the Catholic, 46.9% of the Episcopalian, 47.5% of the public school and 75.9% of the Friends students correctly classified. The composite rankings were then compared with the 13-yr.-old age-group rankings reported by Rokeach.

Segmentation of cervical cell images.

A major problem in the automation of cervical cytology screening is the segmentation of cell images. This paper describes various standard segmentation methods plus one which determines a segmentation threshold based on the stability of the perimeter of the cell as the threshold is varied. As well as contour, certain structural information is used to decide upon the threshold which separates cytoplasm from the background. Once the cytoplasm threshold is found, cytoplasm and nucleus are separated by simple clustering into three groups, cytoplasm, folded cytoplasm and nucleus. These techniques have been tested on 1500 cervical cells that belong to one of eight normal classes and five abnormal classes. A minimum Mahalanobis distance classifier was used to compare results. Manually thresholded cells were classified correctly 66.0% of the time for the 13 class problem and 95.2% of the time on the two (normal-abnormal) class problem. The contour tracing technique was 52.9% and 90.0% correct, respectively.

High resolution analysis of cervical cells--a progress report.

This paper presents preliminary results of research toward the development of a high resolution analysis stage for a dual resolution image processing-based prescreening device for cervical cytology. Experiments using both manual and automatic methods for cell segmentation are described. In both cases, 1500 cervical cells were analyzed and classified as normal or abnormal (dysplastic or malignant) using a minimum Mahalanobis distance classifier with eight subclasses of normal cells, and five subclasses of abnormal cells. With manual segmentation, false positive and false negative error rates of 2.98 and 7.73% were obtained. Similar experiments using automatic cell segmentation methods yielded false positive and false negative error rates of 3.90 and 11.56%, respectively. In both cases, independent training and testing data were used.