Presenting the detection algorithm of all types of vessels in the Persian Gulf region using radar images in order to improve the security of navigation

Detecting the presence of vessels near the coast and far from it is one of the important issues that has many applications in military and civilian industries. This work is very useful for maritime traffic management and traffic safety at sea. Traditional methods of detecting vessels include the use of human observers or systems that transmit characteristics on vessels. New techniques that integrate data from different sources are being developed. In this research, an attempt has been made to optimize the combination of kernels (MKL) by using deep neural network for the first time and finally identify the ship in the Persian Gulf. In this research, Sentinel-1 radar sensor images were used, and finally, using K-means algorithm, 365 training samples of the sea and ships in the Persian Gulf were prepared in different weather conditions. 70% of it is introduced to the network as training data and 30% as test data. In this article, using RBF kernels and polynomials of the first, second and third degrees, features are extracted, and then using a deep neural network, the output of the kernels is combined and high-level features are extracted. The results of the introduced network showed 88% accuracy of the model in identifying and detecting vessels in the Persian Gulf region based on training data. Finally, the network was implemented for validation purposes in Bandar Abbas and Bandar Lange regions and acceptable results were obtained.