A Novel Neighbor Embedding Super Resolution Using Ica and Improved Nlibp
نویسنده
چکیده
In this paper, a novel technique for neighbor embedding single image super resolution (SR) is proposed. An evolutionary algorithm known as Imperialist Competitive Algorithm (ICA) is applied for neighbor embedding single image super resolution. ICA is used for minimizing the reconstruction error of reconstruction weights for neighbors of each low-resolution patch in the low-resolution training image set as a cost function. Additionally, we use improved non-local iterative back-projection (NLIBP) based on edge detection as a post-processing method which perfectly reserves edges in the reconstructed images. The method is applied on various color images and also compared to existing approaches. The results show that the proposed algorithms can more accurately enlarge the low-resolution image than other approaches. The proposed algorithm improves the PSNR by 2.8db on average and reduces the RMSE by 3.2 on average. Visual quality of enlarged images is improved as well.
منابع مشابه
Improved Super-Resolution through Residual Neighbor Embedding
In this paper we first present the machine learning methods applied to the single-image super-resolution issue. Reviewing the novel learning approach of super-resolution through neighbor embedding based on training images, we propose an revised method learning about the residuals from different level, inspired by image pyramids from multiresolution processing. Experiments on gray-level and colo...
متن کاملPso and Ga Based Neighbor Embedding Super Resolution
In this paper a novel technique for Neighbor embedding single image super resolution (SR) is proposed. Given a low-resolution image, its highresolution image is reconstructed from a set of training images, which are composed of one or more lowresolution and corresponding high-resolution image pairs. In this paper we propose a new approach to a single image super-resolution through neighbor embe...
متن کاملPso and Ga Based Neighbor Emedding Super Resolution
In this paper a novel technique for Neighbor embedding single image super resolution (SR) is proposed. Given a low-resolution image, its high-resolution image is reconstructed from a set of training images, which are composed of one or more low-resolution and corresponding highresolution image pairs. In this paper we propose a new approach to a single image super-resolution through neighbor emb...
متن کاملAn Improved Super-Resolution with Manifold Learning and Histogram Matching
Biometric Person Authentication such as face, fingerprint, palmprint and signature depends on the quality of image processing. When it needs to be done under a low-resolution image, the accuracy will be impaired. So how to recover the lost information from downsampled images is important for both authentication and preprocessing. Based on Super-Resolution through Neighbor Embedding algorithm an...
متن کاملInteractive Super-Resolution through Neighbor Embedding
Learning based super-resolution can recover high resolution image with high quality. However, building an interactive learning based super-resolution system for general images is extremely challenging. In this paper, we proposed a novel GPU-based Interactive Super-Resolution system through Neighbor Embedding (ISRNE). Random projection tree (RPtree) with manifold sampling is employed to reduce t...
متن کامل