A fuzzy evidence combination method based on square evidence weight is proposed,and it is used in the dim target multi-feature fusion detection. The basic probability assignment function of the evidence is used to express decision result′s uncertainty. First the detection image′s four features,which are local gray average contrast,local gradient average contrast,local variance and local entropy,are picked up and normalized,then after defuzzification of features,the basic probability assignment of target′s features with supposed set of recognition target can be got according to a priori knowledge and statistical result. After getting basic credibility using adaptive weighted fusion rule it can get the detected target image by decision making rule of game probability distribution. The experimental results show that the method can reduce the uncertainty during the target detection to a large degree and improve the target detection performance of the whole system.