SR-demo is the test function of DELM-SR. SR-demo has four input parameters, which can be introduced in details as follow:
- lambda--------The regularization parameter of the ELM in each leaf nodes.
- NumNodes ---- The number of hidden layer nodes.
- path----------The path of trained models.
- DataName------The testing dateset (Set5, Set14).
- pattern-------The type of test images('png','tmp').
You can run SR_demo by using SR_demo(0.01,200,'model/model_91','Set14','.bmp') as a sample.
You can get the trained models of DELM-SR and DELM-SR+ from BaiduYun
We perform extensive quantitative and qualitative comparisons with current state-of-the-art SR algorithms on 5 datasets: Set5, Set14, BSDS100, Urban100, and Manga109. Set5 and Set14 are included. All the testing_datasets can be download from Link or BaiduYun.
Decision Tree and Extreme Learning Machine base Single Image Super-resolution_temp.pptx is the slide of our method. You can also download it from Here