CAJ | 학술논문

目标检测背景差分法是相对而言较为常用的方法，但是在背景差分法中，背景模型的建立成了至关重要的一个步骤，针对传统的混合高斯模型面临的计算量大、环境尤其是光照突变和对场景构成改变的问题，本文首先提出了一种简化了的混合高斯模型，可以减少运算量[1-3]；其次通过区分是否发生光照突变从而选择不同的背景更新速率。简单、有效地解决了背景建模、实时更新模型参数以及光照变化和场景构成变化的问题。
목표검측배경차분법시상대이언교위상용적방법，단시재배경차분법중，배경모형적건립성료지관중요적일개보취，침대전통적혼합고사모형면림적계산량대、배경우기시광조돌변화대장경구성개변적문제，본문수선제출료일충간화료적혼합고사모형，가이감소운산량[1-3]；기차통과구분시부발생광조돌변종이선택불동적배경경신속솔。간단、유효지해결료배경건모、실시경신모형삼수이급광조변화화장경구성변화적문제。
target detection background difference method is relatively commonly used method, but in the background difference method, background model has become one of the most important step, in view of the traditional gaussian mixture model face a large amount of calculation, the environment, especialy the ilumination mutations and a change to the scene, this paper puts forward an adaptive gaussian mixture model number choice, secondly by distinguishing light mutations and choose a different background update rate. Effectively solve the background modeling, real-time update model parameters and the change of ilumination and scene change.