Research Of Adaptive Threshold Detection Method For Remote Target At Sea From Shipborne Infrared Video

The detection for remote target in the sea-sky background plays a very important role in pirates warning, search and rescue of the ship in distress and vessel collision avoidance. In recent years, the threshold detection method based on the pixel features of the remote target became the focus of remote target detection gradually. But such method sets the threshold or parameter artificial based on the difference of single feature between the target and background, which result in low adaptability and low detection rate in the detection of this method.In order to effectively improve the adaptability and detection rate of the detection method for remote target of shipborne infrared video at sea, this paper carried out a research about aspects of the features described the differences between the pixel of target and background, and the adaptive threshold criterion based on the features is proposed. What’s more,3 methods were proposed based on the research. The first one is a detection method based on the features of remote target at sea with the fundamental idea of the pixel of the small target have characteristics of high gray and gradient simultaneously which is used to distinguish the target pixels between background pixels. Furthermore, the features of the Sea-Sky-Region and the target were used to calculate the thresholds of gray and gradient separately. This method can detect the remote target effectively, but the best detection result needs the artificial adjustment. And the second one could effectively detect the remote target without artificial adjustment by improving the method to calculate the threshold of gradient and filtering clutter, which is called adaptive threshold detection method based on the multi-features of the target. The third method applies the Kmeans algorithm to the target detection on the basis of multi-features of target.A series of test experiments for above methods were carried out to detect the remote target of the infrared video in different sea conditions. The conclusion is that the 3 methods that proposed are effective in the detection of remote target of shipborne infrared video and which have different characteristics on both detection rate and instantaneity.