Proceedings Article | 8 November 2014
KEYWORDS: Image segmentation, Satellites, Remote sensing, Earth observing sensors, Target detection, Satellite imaging, Image processing, Image resolution, Visualization, Binary data
Recently Missing Malaysia Airlines MH370 Flight has attracted worldwide attention. Many countries have worked for the search for MH370, including China, Malaysia, Australia, America, etc. High-spatial resolution satellite remote sensing data has played an important role in searching the lost aircraft. Remote sensing satellite image has advantages on this field, such an having large coverage area and good temporal resolution. The images can provide the information more accurately about disaster, in order for a more rescue. Although remote sensing data has been widely used for earthquake, tsunami, drought and other disasters, the application and research in terms of aviation rescue has been not enough and corresponding searching methods and techniques are not quite mature. The conventional searching methods are mostly based on threshold segmentation and visual interpretation. For emergency rescue, those methods are obviously inefficient, consuming too much time and possibly producing false alarms due to personal negligence and visual fatigue, which bring great disadvantages for locating the crash site and the rescue work in the following.
In this paper, we proposed a new target automatic detecting algorithm base on morphological knowledge for high-spatial resolution remote sensing satellite image. Firstly, we use spectral information from panchromatic high resolution satellite image to segment the image by a threshold; Secondly, according to relationship between the actual size of target and the spatial resolution of image, reduce false alarm rate by morphology algorithm, then the detection result would be obtained; Finally, we can label suspicious districts based on the property of target connectivity and the mutual distance of the targets. These would accelerate the process of locating the target, so as to improve the efficiency of the rescue.
We tested some true satellite images that used in search MH370 airline. The experiment results proved proposed method is practicable. The experiment parts use conventional search methods base on different spectral reflectance between jets and grounds to play threshold segmentation, and then test the proposed method in this paper.
