Research For Ships Detection Technology From Infrared Image In Inland Waterway

As the inland waterway shipping industry developing, the risk of shipping accidents also increased, in particular the relatively low visibility, such as fog weather conditions, the more likely the ship collision incident happens. Therefore the shipping industry is facing a major problem in Ship Collision Avoidance; the key to avoid ship collision is to access the accurate information of vessels. Infrared imaging equipment with anti-interference capability, the climate and environment adaptability, day and night continuous passive detection advantages. In particular, the infrared can overcome the problems that radar can not detect target in the big noise environment. Research on the topic, through installed the infrared imaging system on the ship, capture the video data and using image processing algorithms to detect target ship, provide objective position information of ships to the ship driver, assist the ship driver to make the right decision, reduce ship crash accident, protect people's life and property safety and ensure transport security.This paper firstly studies a variety of image preprocessing algorithms, and then uses the infrared images to test these algorithms, present the results of the algorithms, and analysis the advantage and disadvantage of these algorithms when they are used to process the infrared image.The infrared target ships always appear on the nearby of the Sky-Water Line. In this paper, firstly, detects the Sky-Water Line, and then get the Region of Interest (ROI) through the position of the Sky-Water Line, finally, detect the target ship in the ROI.Aim at the inland River environment, this paper presents a target detection algorithm that based on morphological reconstruction. The principle of morphological reconstruction is to use an image (marker image) to recover another image (mask image) in purpose. Estimate the background by morphological open operator, and then the targets are separated from the background by the Top-hat operator and then make the threshold segmentation, the target finally be completely separated from the background, this result is the mask image. Usually, the largest grayscale value of pixels region in the Infrared image is the chimney, engine or other higher temperature part of ship, the paper selects this region as marker image. Finally, use the reconstruction to detect the infrared target. In addition, this paper optimizes the gray morphology algorithm, and greatly improves the performance of the algorithm, and makes sure the algorithm can be used in real-time application.Overall, the accuracy and run speed of the whole target detection algorithm of this paper are expected to achieve the goal of research.