• Communications for Statistical Applications and Methods
• /
• v.10 no.1
• /
• pp.225-232
• /
• 2003

The support vector machine (SVM) is becoming increasingly popular in classification. The import vector machine (IVM) has been introduced for its advantages over SMV. This paper tries to improve the IVM. The proposed method, which is referred to as the polychotomous machine (PM), uses the Newton-Raphson method to find estimates of coefficients, and the Rao and Wald tests, respectively, for addition and deletion of import points. Because the PM basically follows the same addition step and adopts the deletion step, it uses, typically, less import vectors than the IVM without loosing accuracy. Simulated and real data sets are used to illustrate the performance of the proposed method.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.479-481
• /
• 2003

Large number of coral reefs in Thailand waters make the use of satellite imagery probably the only practical method for their monitoring. This paper reports the result of detecting small shallow-water coral reef by using maximum likelihood classification technique. Combination of blue/green and near-infrared band ratio are used as spectral signatures derived from a Landsat 7 imagery covering western portion of the Gulf of Thailand. Result assessment reveals accuracy significantly over 60 percent. The result is encouraging and would be a basis for further study to realize the full potential and limitation of this technique.

• Korean Journal of Remote Sensing
• /
• v.7 no.2
• /
• pp.149-163
• /
• 1991

This paper discusses the capabilities of airborne remotely sensed data to detect and classify forest damades. In this work the AMS (Aircraft Multiband Scanner) was used to obtain digital imagery at 300m altitude for forest damage inventory in the Black Forest of Germany. MSS(Multispectral Scanner) digital numbers were converted to spectral emittance and radiance values in 8 spectral bands from the visible to the thermal infrared and submitted to a maximum-likelihood classification for : (1) tree species ; and. (2) damage classes. As expected, the resulted, the results of MSS data with high spatial resolution 0.75m$\times$0.75m enabled the detection and identification of single trees with different damages and were nearly equivalent to the truth information of ground checked data.

• Journal of Forest and Environmental Science
• /
• v.37 no.1
• /
• pp.44-51
• /
• 2021

Rapid urbanization and population growth have caused substantial land use land cover (LULC) change in the Kathmandu valley. The lack of temporal and geographical data regarding LULC in the middle mountain region like Kathmandu has been challenging to assess the changes that have occurred. The purpose of this study is to investigate the changes in LULC in Chandragiri Municipality between 1996 and 2017 using geographical information system (GIS) and remote sensing. Using Landsat imageries of 1996 and 2017, this study analyzed the LULC change over 21 years. The images were classified using the Maximum Likelihood classification method and post classified using the change detection technique in GIS. The result shows that severe land cover changes have occurred in the Forest (11.63%), Built-up areas (3.68%), Agriculture (-11.26%), Shrubland (-0.15%), and Bareland (-3.91%) in the region from 1996 to 2017. This paper highlights the use of GIS and remote sensing in understanding the changes in LULC in the south-west part of Kathmandu valley.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.5 no.1
• /
• pp.27-37
• /
• 2002

The classification accuracy of land cover has been considered as one of the major issues to estimate pollution loads generated from diffuse landuse patterns in a watershed. This research aimed to assess the effects of the acquisition methods and sampling size of training reference data on the classification accuracy of land cover using an imagery acquired by optical sensor(OPS) on JERS-1. Two kinds of data acquisition methods were considered to prepare training data. The first was to assign a certain land cover type to a specific pixel based on the researchers subjective discriminating capacity about current land use and the second was attributed to an aerial photograph incorporated with digital maps with GIS. Three different sizes of samples, 0.3%, 0.5%, and 1.0% of all pixels, were applied to examine the consistency of the classified land cover with the training data of corresponding pixels. Maximum likelihood scheme was applied to classify the land use patterns of JERS-1 imagery. Classification run applying an aerial photograph achieved 18 % higher consistency with the training data than the run applying the researchers subjective discriminating capacity. Regarding the sample size, it was proposed that the size of training area should be selected at least over 1% of all of the pixels in the study area in order to obtain the accuracy with 95% for JERS-1 satellite imagery on a typical small-to-medium-size urbanized area.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.14 no.1
• /
• pp.39-47
• /
• 1996

This paper is to provide land cover classification techniques for large area ranged in different pathos by classifying Landsat TM data of Jeonnam province. The analyses proceeded by individual scene because acquired dates are not same in different pathes. In this processing, troubles had happened something like variation of classes can be classified in two scenes and choice problem about overlapped area. Since spatial effects in large area affect data values, it was difficult to make a selection of classes and training fields. we could present a solution about these problems by trial and error method, and found that Bayesian maximum likelihood classification and majority filtering were effective to improve classification accuracy.

• Proceedings of the KSRS Conference
• /
• 2007.10a
• /
• pp.6-10
• /
• 2007

Classification of hyperspectral images is challenging. A very high dimensional input space requires an exponentially large amount of data to adequately and reliably represent the classes in that space. In other words in order to obtain statistically reliable classification results, the number of necessary training samples increases exponentially as the number of spectral bands increases. However, in many situations, acquisition of the large number of training samples for these high-dimensional datasets may not be so easy. This problem can be overcome by using multiple classifiers. In this paper we compared the effectiveness of two approaches for creating multiple classifiers, feature selection and feature extraction. The methods are based on generating multiple feature subsets by running feature selection or feature extraction algorithm several times, each time for discrimination of one of the classes from the rest. A maximum likelihood classifier is applied on each of the obtained feature subsets and finally a combination scheme was used to combine the outputs of individual classifiers. Experimental results show the effectiveness of feature extraction algorithm for generating multiple classifiers.

• IEIE Transactions on Smart Processing and Computing
• /
• v.4 no.4
• /
• pp.202-208
• /
• 2015

In this study, we propose a new inference algorithm for a multiclass Gaussian process classification model using a variational EM framework and the Laplace approximation (LA) technique. This is performed in two steps, called expectation and maximization. First, in the expectation step (E-step), using Bayes' theorem and the LA technique, we derive the approximate posterior distribution of the latent function, indicating the possibility that each observation belongs to a certain class in the Gaussian process classification model. In the maximization step, we compute the maximum likelihood estimators for hyper-parameters of a covariance matrix necessary to define the prior distribution of the latent function by using the posterior distribution derived in the E-step. These steps iteratively repeat until a convergence condition is satisfied. Moreover, we conducted the experiments by using synthetic data and Iris data in order to verify the performance of the proposed algorithm. Experimental results reveal that the proposed algorithm shows good performance on these datasets.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.28 no.5
• /
• pp.493-504
• /
• 2010

This study proposes a classification method based on an automated training extraction procedure that may be used with very high resolution (VHR) images of non-accessible areas. The proposed method overcomes the problem of scale difference between VHR images and geographic information system (GIS) data through filtering and use of a Landsat image. In order to automate maximum likelihood classification (MLC), GIS data were used as an input to the MLC of a Landsat image, and a binary edge and a normalized difference vegetation index (NDVI) were used to increase the purity of the training samples. We identified the thresholds of an NDVI and binary edge appropriate to obtain pure samples of each class. The proposed method was then applied to QuickBird and SPOT-5 images. In order to validate the method, visual interpretation and quantitative assessment of the results were compared with products of a manual method. The results showed that the proposed method could classify VHR images and efficiently update GIS data.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.34C no.9
• /
• pp.61-67
• /
• 1997

The HMM-Net is an architecture for a neural network that implements a hidden markov model(HMM). The architecture is developed for the purpose of combining the classification power of neural networks with the time-domain modeling capability of HMMs. Criteria which are used for learning HMM_Net classifiers are maximum likelihood(ML), maximum mutual information (MMI), and minimization of mean squared error(MMSE). In this classifiers trained by the gradient descent algorithm with the above criteria. Experimental results for the isolated numbers from /young/to/koo/ show that in the binary inputs the performance of MMSE is better than the others, while in the fuzzy inputs the performance of MMI is better than the others.