Hyperspectral remote sensing image has high-resolution spatial information and contains continuous spectrum information,and it has unique advantage in the field of target detection.Target detection techniques based on hyperspectral image are one of the important areas in remote sensing theory and application.Starting from the concept of correlation coefficient in statistics,the target detection algorithm of hyperspectral image based on constrained maximum correlation coefficient is proposed.Different amounts of sub-pixel targets and goals were gotten by adding the target spectrum in the real hyperspectral image,and the real hyperspectral images of large targets were obtained by hyperspectral imager.By comparing constrained energy minimum method and the proposed method,the experiment results show that the proposed algorithm has more robust performance for large target detection.