Complex Moving Target Detection In The Star Field And Bilateral Filtering Algorithm

With increasing development of space flight techniques, it is becoming an important issue to detect and extract complex moving objects in moderate and near distances from image sequence obtained by a moving space-based camera. Owing to the movement of the camera carrier, both the star field and objects are moving and this double movement makes it difficult to detect and extract objects from image sequence. This paper presents a method to detect and extract complex objects from star field. First, based upon the characteristics of the star field and complex objects, we propose a registration method of the star field background based on isolated star groups; and then, the moving objects are detected from difference images of the registered image sequence with the region-growing method used to extract the contours of the complex objects; finally, aiming at eliminating the potential fictitious objections resulting from proliferation between frames, we make most of the outcome images obtained from the previous two steps with some operation done. The simulated experiments show that the proposed method can effectively and efficiently detect and extract complex objects from the star field.Image denoising is critical and indispensable for image processing. There exist many methods for this task, including iterative and non-iterative methods, linear and nonlinear methods. Bilateral filtering can be regarded as a non-iterative and nonlinear method for image denoising. The bilateral filtering method is very effective in preserving edges and textures when noise is suppressed. The most critical issue in the bilateral filtering is the parameter selection of filters, which is highly related with the denoising performance. Different from the fixed parameter filter in the traditional bilateral filtering, we propose an adaptive parameter selection method, in which the parameters of the filters match the local statistics of the image to be processed. The experimental results show that the proposed method is more effective than the traditional bilateral filtering method.