Research On Image Segmentation Under Complex Background

The imaging guidance technology has been widely applied in modern warfare, but the maximum detection range is critical. The basic problem inherent to extent the detection range is the detection of small, low observable targets in images and in some sense is to find an efficient method for increasing operating range. The difficulties in low contrast image segmentation lie in the follows: the distance is far,the area is small and the object often submerges in the background.Therefore, comparing with other target segemetations,how the targets can be robustly detected under complex background has become a more realistic and challenging research.In order to solve this problem, we conducted the following studies.In this paper, we have sumed up and analyzed the conventional background estimation techniques, and finally the approach based on the adaptive gray-scale morphological has been used to reduce the background, enhance the target information.For the imaging size changes along with the distance, the key point of the algorithms is to determine a kind of structuring element which can change adaptively. A new method combined with the motion information of the sensor have been put forward.Because there is no wave crest in histogram for low contrast image, the classical methods are not adapted to them.The new method has been improved on segmentation based on two-dimensional maximum entropy have been presented combining with the threshold segmentation and the characteristic of the gray-scale distribution concentrated, and a simplified algorithm is used to further reduce operation time.The experimental results indicate that the simplified algorithm can decrease the time consumption effectively, have good real-time capability and easy to engineering.It basically solved the problem of segmentation in the complex background.