Based On The High Score Images Mining Information Extraction

Mine resource is the important foundation of the national economic development,but the prohibition illegal mining.How to fast and efficient extraction of mine information governance is of great importance to the illegal mines.Mining information extraction using artificial visual interpretation,human-computer interaction,such as the traditional method,however,people in recent years has been committed to research of the remote sensing image automatic segmentation.But more research is one of the traditional classification method.So this paper mainly for mining information extracted quickly and conduct research.In jilin province key mining area as the study area of the information extraction.Introduction on the study area of jilin province in the first place.Then this paper USES four methods to classify remote sensing image processing,respectively is the traditional maximum likelihood supervised classification method,the traditional ISODATA unsupervised classification method and decision tree classification decision tree classification method and mean shift and simply expounds the principle and concept of four methods.The classification results of four methods used error matrix to analysis and comparison,it is concluded that the precision of four types of methods are: mean shift decision tree classification overall accuracy is 87.56%,Kappa coefficient is 0.8267,mining classification accuracy is 76.88%;The decision tree classification overall accuracy is 80.58%,Kappa coefficient is 0.6661,mining classification accuracy is71.27%;Maximum likelihood classification overall accuracy is 70.90%,the Kappa coefficient is0.6012,mining classification accuracy is 65.77%;ISODATA classification of overall accuracy is60.3849%,Kappa coefficient is 0.4566,mining classification accuracy of 41.86%.It is concluded that the high score four image classification method,the mean shift decision tree classification effect is best,overall classification accuracy,Kappa supreme,mining classification accuracy are the highest.Based on the above results can be obtained in mine information extraction based on mean shift algorithm has certain feasibility.The algorithm of mining information extracting provides a new way of thinking.