An adaptively-weighted blind image restoration algorithm based on energy constraint is proposed. The images are divided into several sub-images and gradients of sub-images are introduced as weights to build the estimation model of weighted optical transfer function, which can reduce the influence of image texture on the estimation of optical transfer function. Based on the energy of image signals, the constraint equation is established, and the optimal restoration result is chosen by the dichotomy to realize adaptive blind image restoration. Results of simulation and multispectral remote sensing image experiments show that the proposed algorithm can produce high peak signal-to-noise ratio and structural similarity, which will effectively restore Gaussian blurred images, enhance the image resolution, and improve subjective visual effects. The proposed algorithm can be applied to the fields requiring large data and real-time monitoring.