TM Classification Research Articles

Classification by Re-Estimating Statistical Parameters Based on the Autoregressive Model

To investigate the effect of pixel-to-pixel correlation on classification performance for TM data, some experiments were conducted. One result was that the pixel-to-pixel correlation in homogeneous fields of TM data is larger than that of the simultaneously acquired MSS data, a finding that implies it is more important to consider the pixel correlation for TM classification rather than MSS classification.To overcome the effect of the correlation, the following methods were investigated. Based on the autoregressive model, the Gaussian statistical parameters for the correlated image were modified to those for uncorrelated stochastic process. An example using agricultural TM data shows a slight improvement in terms of the weighted mean percentage of correct classification. By investigating the correlation structures, a more realistic model for TM data was realized.

Optimum threshold of maximum likelihood decision rule for TM classification.

Optimum threshold of maximum likelihood decision rule for TM classification.

画素間相関を考慮したＴＭ画像の最優法分類

To investigate effects of pixel to pixel correlation on classification performance for TM data, some experiments have been conducted. One of the results shows that pixel to pixel correlation in homogeneous fields of TM data is larger than that of the simultaneously acquired MSS data. This implies that it is more important to take into account the pixel to pixel correlation for TM classification rather than MSS classification.To overcome the effect due to the correlation, the following methods have been investigated.Based on the auto-regessive model, modify the Gaussian statistical parameters for the correlated image to those for uncorrelated stochastic process.An example using an agricultural TM data shows a slight improvement in terms of the weighted mean percentage of correct classification. Through the investigation of the correlation structures, more realistic model for TM data has been clarified.

Landsat classification of Argentina summer crops

A Landsat MSS and TM classification approach based on three features derived from the greenness profile has proved very effective in separating and identifying corn, soybeans, and other ground cover classes in the United States. The objective of this study is to investigate the separation of summer crops in Argentina, one of the most important commodity exporters, using the same greenness profile features that have proved effective in the U.S. Corn Belt. The area chosen for study is a more complex cropping practice area located in the north-west corner of Buenos Aires province in Pampa Humeda, where corn, soybean, sorghum, sunflower, and pastures are cultivated. It is shown that the profile features can provide very effective separation, except in the case of corn from sorghum. Separation between corn and soybeans was found to be greater than in the United States. This study suggests that the automatic, unsupervised classification approach developed in the United States, with relatively minor modification, can be used for summer crop area estimation in Argentina.

Purification of Training Samples for Supervised TM Classification

A methodology for purification of training samples for the pixel-wise Maximum Likelihood Classification is proposed. In this method, pixels which show comparatively high local spectral variability as well as spectrally separable classes are removed from the preliminary designated training samples. An example using agricultural TM data shows that separability can be improved 3.78 times in terms of divergence between a specific class pair, goodness of fit to Gaussian can be improved 0.14 times in terms of Chi-square, 11.9%improvement of the weighted mean percentage classification accuracy can be achieved, and most importantly a 20.6%improvement of probability of correct classification can be achieved for a specific class.

Effect of Landsat thematic mapper sensor parameters on land cover classification

Selected sensor parameter differences between TM and MSS were assessed through classification performance of a suburban/regional test site. Overall classification accuracy of a seven-band Landsat TM scene in comparison to MSS yielded an improvement in accuracy from 74.8% to 83.2%. To study the possible causes for the difference in classification performance, key sensor parameter differences between MSS and TM, including 1) spatial resolution (30 m for TM versus 80 m for MSS), 2) quantization level (256 levels for TM versus 64 for MSS), and 3) spectral regions (seven bands in four major spectral regions for TM versus four bands in two regions for MSS), were evaluated. Landsat TM data were processed to simulate all possible combinations of these MSS and TM parameters, yielding a three-factor design with two levels per factor. The results indicated that the added spectral regions (TM 1, TM 5, and TM 7) and to a lesser degree the increase in quantization level to eight bits produced the improved TM classification accuracy. However, in this study, the higher 30 m spatial resolution of TM contributed to a reduced classification accuracy from increased within-field variability or class heterogeneity.