Abstract: A learning-based super-resolution algorithm based on Kernel Partial Least Squares (KPLS) regression is proposed. First, KPLS regression algorithm is introduced. Then a super-resolution algorithm based on KPLS regression is analyzed. High resolution images use the high-frequency information as their feature, while low resolution images use middle-frequency as their features. Based on the relationship of the high and low resolution images, KPLS is used to set up regression model. The regression model is applied to infer high-resolution image. The experimental results show that our method can achieve very good results to face images and car plate images. The results of our method are closer to the real images.