论文部分内容阅读
利用自然背景与人造目标在分形维数上的差异,提出了一种针对自然背景下的快速红外目标检测技术。通过对图像各个子图的改进不规则分形维数的计算,得到各个子图的指标函数J曲线;经过对J曲线的适当策略分析,可以得到目标最大可能存在的子图区域,或者多目标可能存在的区域。仿真实验证明:算法可以快速检测出红外图像中人造目标可能出现的区域,为随后的识别跟踪等分析降低了计算量。
Based on the difference of fractal dimension between natural background and man-made target, a fast infrared target detection technology against natural background was proposed. By calculating the improved fractal dimension of each subgraph of the image, the J curve of the index function of each subgraph is obtained. After the appropriate strategy analysis of the J curve, the maximum possible subgraph region of the target can be obtained, or the multi-target probability Existing area. The simulation results show that the algorithm can quickly detect the possible areas of artificial targets in infrared images and reduce the computational load for the subsequent identification and tracking.