P Zeaiean Firozabadi; P Safarbeyranvand; A Hosseinhgholizade; R Hasanitabar; M Safarbeyranvand
Volume 10, Issue 1 , June 2018, , Pages 1-16
Abstract
The issue of the mapping of rock units in an ever-expanding area has now reached a point where the detection and classification of rock units is carried out through the aid of hyperspectral image. In this research, Hyperion images are used in the light of the work of previous researchers and the application ...
Read More
The issue of the mapping of rock units in an ever-expanding area has now reached a point where the detection and classification of rock units is carried out through the aid of hyperspectral image. In this research, Hyperion images are used in the light of the work of previous researchers and the application of the SAM supervised classification algorithm for the detection and separation of rock units in Khorramabad region, Lorestan province. After performing the necessary preprocesses including atmospheric correction performed by the FLAASH method, linear MNF transformation was used to determine the dimension of main image, to separate the noise from other information and reduce the processing level in the next stages, and the PPI algorithm to find the pixels that More purity is used in multispectral images. From the overlapping of pure pixels with rock units and based on ground data from the study area, the average range was extracted for each member. Then, these pure members were used as inputs for the above-mentioned algorithms and image categorization was used. Finally, the mapped classification of this method was compared with existing maps and land data and their accuracy was checked. The accuracy of the SAM method was verified by verifying the accuracy of the algorithm by calculating the error matrix with the highest of 68.83% and kappa coefficient of 0.49%, which indicates the importance of hyperspectral images and the SAM method in separating the rock units.