One robust method of tracking targets in infrared imagery was proposed for detecting the hidden target that has the characteristics of being hard to be found, complex, changeable and less information, etc. About target detection, fuzzy C-means clustering was used to cluster the regions in the image firstly and the candidate target area was segmented according to the brightness information preliminary, then further segmentation was made by texture energy. At last, the target area was detected. As for the hidden target with no consistent contour, the chain code method was used to extract the target contour, and target was tracked through centroid tracking algorithm. The experiment results show that the method is effective for occlusion target detection and tracking.%针对隐蔽目标难于发现、复杂多变、信息量少等特点,提出了一种用红外热像仪检测与跟踪遮挡目标的方法。关于目标检测,首先利用模糊 C 均值聚类(fuzzy C-means clustering, FCM)对图像中的区域进行聚类,根据亮度信息初步分割出候选目标区域,然后用纹理信息进一步分割候选的目标区域,以检测出最终的目标区域;关于目标跟踪,由于被遮挡的目标轮廓不连贯,所以采用链码提取目标轮廓,并应用质心跟踪算法跟踪目标。通过实验,验证了该方法对遮挡目标检测和跟踪的有效性。
展开▼