LAUSR

The wide-field surveillance camera plays a critical role in space debris detection and the visible light situational awareness required for early warning. However, stray light in such a system has always been a serious problem. Current methods cannot effectively eliminate the interference of stray light, which directly lead to the inability to accurately segment the target and background, which greatly reduce the accuracy of target recognition. To solve this problem, we proposed an accurate stray light elimination method based on recursion multi-scale gray-scale morphology (RMGM). First, we defined two structural operators with different domains. These two structural operators can make full use of the difference information between the target region and the surrounding background region, which is the basic premise to ensure high-precision correction. Then we used the two structural operators with different domains to perform morphological processing on the surveillance image to eliminate stray light. Finally, in order to ensure that targets with different sizes in the surveillance image are not lost, we adopt a recursion multi-scale method. We increase the size of structural operator and perform the morphological operation again on the estimated and eliminated stray light non-uniform background in order to retrieve the lost larger size target. In addition, we add an automatic decision mechanism on the recursion multi-scale method by using corresponding threshold judgment. Further experimental results on real captured image datasets show that compared with other methods, the proposed RMGM method can simultaneously have high-precision stray light elimination effect, high-precision target retention rate, and faster computation time.