Building Image Feature Extraction Using Data Mining Technology

Abstract

At present, data mining technology is continuously researched in science and application. With the rapid development of remote sensing satellite industry, especially the launch of remote sensing satellites with high-resolution sensors, the amount of information obtained from remote sensing images has increased dramatically, which has largely promoted the application of remote sensing data in various industries. This technique mines useable information from less complete and accurate data while ensuring low program complexity. In order to determine the impact of data mining techniques on feature extraction of graphic images, this paper explores the relevant steps in the image recognition process, especially the image preenhancement and image extraction processes. This paper develops a preliminary set of relevant data and investigates two different extraction methods based on the availability or absence of nursing information. Aiming at the advantages and disadvantages of the two house extraction methods, this work discusses how to effectively integrate remote sensing data. It uses different data sources to describe different characteristics of buildings, analyzes and extracts effective information, and finally derives building information. The research results show that, using the SVM algorithm in data mining for image feature extraction, in the verified filtering window, the accuracy can be effectively improved by about 20%. Buildings are important objects in high-resolution remote sensing images, and their feature extraction and recognition technology is of great significance in many fields such as digital city construction, urban planning, and military reconnaissance.