Use of Antcolony algorithm to Improve Classification accuracy Case study: TM Sensor

Today, remote sensing images are known as the latest information to study land cover and land use.The position of the title in this section is ۱۲۰ mm from the top of the page or upper edge. for this purpose used of TM sensor data of Landsat satellite. Pre-processing operations were performed on the satellite image and the processing stage included the implementation of educational samples on the image and image classification using maximum probability and neural network methods in order to extract thematic land use maps. After obtaining the error matrix related to the two methods, the accuracy and kappa coefficient were obtained for them. The results show that the neural network method has better classification accuracy than the maximum probability method. Based on the error matrix obtained from the classification, the kappa coefficient for the maximum probability method was ۰.۸۰۲۲ and its overall accuracy was ۸۳.۵۵% and the neural network method respectively was ۰.۸۴۳۵ and ۸۸.۴۵%. Finally, to improve the accuracy of image classification, the ant colony algorithm has been used, which is a new way to improve the accuracy of supervised classification of remote sensing images with limited educational data. This algorithm was programmed in MATLAB software and after classifying the images using this program, the accuracy of the classification for the TM sensor with this algorithm in the classification of the maximum probability is about ۴.۸% and for the classification of the neural network to Approximately ۱.۶۵ percent increased.

Keywords:

Classification , Optimization , Nueral Network , Maximum Likelihood , Ant Colony