论文部分内容阅读
自然背景和人造目标对于某些分形特征存在着一定的本质差别,因此充分利用这些差别,能够为目标探测提供一套全新的方法。从仿真实验的处理结果来看,这种方法对于自然背景中嵌入少量人造目标这类简单情况的检测效果较好,且具有较强的抗噪声干扰能力;对于其他复杂情况,本文分别采用阈值分割组合法和K均值聚类法进行尝试,得到了一些初步结果。
Natural backgrounds and man-made objects have certain essential differences for some fractal features, so making full use of these differences can provide a completely new approach to target detection. From the results of simulation experiments, this method has a good effect on detecting simple cases with a small number of man-made objects embedded in the natural background and has strong anti-noise ability. For other complicated cases, the threshold segmentation Combination method and K-means clustering method, some preliminary results are obtained.