Development of a New Method for Edge Detection from High-Resolution Aerial/Satellite Images, with Emphasis on Threshold Optimization and Using Imperialist Competitive Algorithm

Edges are one of the salient properties of the image in which they have important information of the image and represent shape characteristics of the objects. Edges are important features due to the fact that the human visual system uses a preprocessing step for edge detection. The majority of the classical mathematical algorithms for the edge detection، such as Gradient، Laplacian and Laplacian of Gaussian operators are based on the derivative of the original image pixels. In the remote-sensing imagery، because of the high rate of changes، these edge detection operators perform weakly in correct detection of the feature boundaries and keeping their consistency. In order to solve these problems، this research presents a new technique using Shannon entropy based on Imperialist Competitive Algorithm (ICA). In this method، firstly a piecewise thresholding is used to identify the threshold of different parts of the image، and then the area boundaries are extracted using the Shannon entropy based on the selected threshold. According to obtained results، selection of thresholds have a high influence on the final results، Based on this، ICA optimized method is used in this research. In order to evaluate the performance of algorithm، the results from the proposed technique are also compared with the results obtained from Canny، LOG، Sobel، Roberts، Ant colony optimization edge detector and other Entropy edge detectors. The results show that the proposed method presents higher reliability in detecting the edges of different digital images.