Research On Typical Ground Target Detection Technique In Infrared Image Guidance

Infrared image guidance is intelligent and precise guidance, which uses the difference of radiation between the target and the background to get the infrared images. The infrared image is processed real-timely by the high speed processor, making the infrared image guidance an efficiency guidance method. The processing technology of infrared image information is one of the pivotal taches in the infrared image guidance system. Under the modern high technic condition, the battlefield rivalry becomes vehemence increasingly, which requires the infrared guidance weapon be capable of detecting the target in the nature background and artificial interference.The ground targets, such as airport runway, the road, the dam, the early-warning radar, the electricity factory, the oil storeroom and so on, have gradually become the first-selected targets to destroy. However, for the reason that the background of the ground targets are usually comparatively complex, as well as the too many kind of ground substance and the high degree influence on the target characteristics by the weather, it is not only tarnally difficult but also very defiantly to detect the targets mentioned above, which makes it be a research hotpoint in the target recognition field.Based on the task background of automatic target recognition in infrared image guidance, this thesis carried out a thorough study on the detection of two kind of typical targets, including the airport runway and circle small dimensionality target in the infrared image guidance.This thesis established a simplified airport runway model based on the analysis about the infrared radiation characteristics of the airport runway. We made a study on three kind of runway target detection algorithm using the edge-line abstraction, and then a new runway target detection algorithm based on the cross-projection width spectrum (CPWS) was proposed. Firstly, we creatively introduced the concept of CPWS, an analysis about the characteristics of which was made subsequently. Then an integrated decision was made with the D-S evidence theory, which enhanced the peak value characteristic of CPWS, to implement the abstraction of the runway edge points. Finally a sequence least square algorithm under the distance constraint was proposed to carry through the line fitting with the target edge points abstracted. The experiment results indicated the good detection accuracy and real-time performance of the algorithm proposed above.Base on an integrated study about the edge-based methods at present for the circle small dimensionality targets detection, this thesis proposed a new algorithm based on the edge-tracking and the cumulating extremum in the parameter space for the small dimensionality targets detection. By the edge tracking under constraints, a great deal of interference was eliminated, and then points picked out with equal interval from subsections of the edge-lines were used to calculate the circle parameters, which reduced the inefficacy cumulating enormously. The experiment results verified the efficiency and real-time performance of our algorithm.