针对单光谱图像用于目标检测时信息量不足的缺点，提出了一种可见光图像与红外图像基于压缩域融合检测的方法。该方法首先使用合适的模型构造目标原子，得到超完备字典，再对待测图像在字典上分解所得稀疏系数进行融合，最后通过稀疏度指标对融合系数进行判定，得到目标所在位置。实验结果表明，与单帧图像检测方法相比，该方法使得待检测图像信息更加丰富，提高目标的检测率。

A target detection approach is developed using compressive fused image for target detection. Firstly, an over-complete dictionary is constructed with atoms which are produced by two-dimensional Gaussian model. Secondly, we encode the sensor data on the constructed over-complete dictionary and combine the coefficients with the fusion impact factor. Targets can be determined by the sparse index of the fused coefficients. Experiment results show that the proposed approach has a higher recognition rate on account of the information enhanced.