Efficient Super Resolution Algorithm using Overlapping Bicubic Interpolation
Abstract : In practical CCTV applications, there are problems of the camera with low resolution, camera fields of view, and lighting environments. These could degrade the image quality and it is difficult to extract useful information for further processing. Super-resolution techniques have been proposed widely by the researchers. However, many approaches are complex and are difficult to use in practical scenarios. In this paper, we propose an efficient Super-resolution algorithm using overlapping bicubic for hardware implementation. Experimental results are verified using processing time and reconstructed images that can be used in real time applications.
? The major advantage of the super resolution approach is that it may cost less and the existing LR imaging systems can be still utilized.
? This method does not interpolate values, it just copies existing values. For 2-D, the number of grid points required to evaluate the interpolation function is four.
? It is our understanding that most of the existing SR approaches in their default settings take the downsampled version of the HR images as the LR image inputs in their training.
? The message is that material scientists cannot simply grab an existing pre-trained SR model for processing the paired EM images.
? This pair-specific information does not exist in the general single-image SR when an external training set is used.
? The problem, embedded video super-resolution approach can be implemented and connected directly to CCTV.
? The quality of the single image interpolation is very much limited due to the ill-posed nature of the problem, and the lost frequency components cannot be recovered.
? In the SR setting, however, multiple LR observations are available for reconstruction, making the problem better constrained.
? The basic problem is to obtain an HR image from multiple LR images.
? The basic assumption for increasing the spatial resolution in SR techniques is the availability of multiple LR images captured from the same scene.
• The proposed approach is based on alternating minimization (AM) of a new cost function with respect to the unknown high-resolution (HR) image and blurs, which improves the quality of blur.
• A hierarchical idea of image fusion has been proposed for combining significant information from several images into one image.
• Registration is the process of bringing all the shifted versions of low resolution images into a single plane with respect to a reference image.
• Feature based registration comprises of feature detection, feature matching, optimum transformation and up-sampling and provides better results in many applications.
? Bicubic interpolation was firstly proposed in with several variations to increase the performance.
? A closed-circuit television (CCTV) is used for urban security and environmental monitoring.
? In the set-up of smart cities, home video surveillance system provides the necessary visual information to improve situational awareness, to enhance security, to deter possible theft, and to provide remote monitoring.
? The advantages of the LB-NLM method are its simple model structure and short training time, while its performance is less accurate only by a small margin than the deep learning based SR methods.
|