CAJ | 학술논문

针对舰载红外警戒系统目标图像中弱小目标的检测问题,提出了一种基于离群点挖掘的目标检测算法(OBTD).依据目标与背景在灰度分布上的不同特性,算法将弱小目标视为离群点.分割图像为一系列的子图像,计算子图像中各像素值的局部可达密度,设定阈值挖掘出离群点,得到可能的目标点.算法时间复杂度低,实时性好.试验表明,该算法能有效地检测出图像中的弱小目标,便于进一步的目标识别和跟踪.
침대함재홍외경계계통목표도상중약소목표적검측문제,제출료일충기우리군점알굴적목표검측산법(OBTD).의거목표여배경재회도분포상적불동특성,산법장약소목표시위리군점.분할도상위일계렬적자도상,계산자도상중각상소치적국부가체밀도,설정역치알굴출리군점,득도가능적목표점.산법시간복잡도저,실시성호.시험표명,해산법능유효지검측출도상중적약소목표,편우진일보적목표식별화근종.
To resolve the problem of the dim targets detection for shipboard infrared search and track system target image,an algorithm called outlier-based targets detection(OBTD)is presented.According to the different characteristic between targets and background,dim target can be considered as outlier.Divide the origin image into small images;Calculate the local-reachable-density for every pixel;after that,the outlier can be found by setting threshold and the possible targets are obtained.The algorithm merit lie in a good real-time because of a low time complexity.The experiment shows that the algorithm can detect the dim target effectively,and it can offers convenience to further target recognition and track.