Detection, classification and postclassification of urban features from multispectral images and mobile LIDAR point clouds

Abstract

The scientific and technological development that happened in the second half of the twentieth century led to new tools, techniques, and technologies that forced many scientific disciplines to renovate, introducing and adapting these advances to the classical techniques to meet the new needs of society. The development of optics and sensors capable of taking information from various regions of the electromagnetic spectrum, the launching of satellites, and advances such as laser systems are contributions that have gradually been included in remote sensing works. Recent times have highlighted the importance of pattern recognition techniques in classification procedures and information extraction tasks in engineering, computer science, or mathematics issues. To certain sciences, such as remote sensing, these techniques are particularly important because they allow automatic cartographic entities detection and classification processes. Carrying out these procedures manually would be too expensive and time-consuming because of the high volume of remotely sensed information currently available