RESEARCH ARTICLE Study of the Behavior of Super Resolution on Soft-Classified Output Shubham Rana 1 • Anil Kumar 1 Received: 17 October 2018 / Accepted: 17 July 2019 Ó Indian Society of Remote Sensing 2019 Abstract Once the satellite sensor is in orbit, no hardware enhancement of the lens assembly can be done to improve spatial and spectral resolution. Super resolution (SR) as single frame or multi-frame can solve this problem up to a large extent. In this study, single- and multi-frame SR techniques were applied and tested on Worldview-2 datasets as well as on across-spatial datasets of LISS III and LISS IV. Study of soft classifier’s behavior on super-resolved images was performed through possibilistic C-means classifier. Quantitative methods based on calculation of peak signal-to-noise ratio, mean square error, root means square error, image quality index and qualitative methods of visual interpretation proved that both super- resolution methods remove outliers in an efficient way and resulted in images containing sharp edges. The single-frame super-resolution technique was found relatively inferior in terms of contrast and spatial resolution. Overall, multi-frame SR method outperformed other methods. Keywords SF—single frame Á MF—multi-frame Á SR—super resolution Á HR—high resolution Á LR—low resolution Introduction The paramount objective of super resolution (SR) is to synthesize an image of higher resolution from one or multiple images of lower resolution. Apart from higher entropy levels, increased pixel density is another significant characteristic of higher-resolution images as compared to the lower-resolution images. Therefore, higher the resolu- tion more will be the information content in the original image frame. The requirement of high-resolution imaging is quite prevalent in computer vision applications so as to have an effective performance in applications pertaining to pattern recognition and object detection (Capel and Zis- serman 2003). Many applications require magnification or enlargement of a particular area of interest in the image, where high resolution becomes a dire need, for example surveillance, forensics and satellite imaging. The first and foremost prerequisite of super resolution is the availability of multiple low-resolution image frames of the original frame, in order to augment the spatial resolution through SR techniques. Altogether, the basic fundamental behind all SR techniques is restoration of an HR image from multiple aliased or noisy LR images. Problems related to SR techniques are registration, restoration and interpolation of images. Image restoration removes the additive white Gaussian noise as well as blurredness from the image, but it does affect the aspect ratio (dimensionality) of the image. Qualitatively and theoretically, image restoration may seem quite similar to SR reconstruction. However, SR reconstruction was believed to be an improvised second- generation alternative to image restoration. There are sev- eral factors for poor spatial resolution in the remote sensing data (Cho et al. 2007). Few of them are as follows: large altitude of the sensor system, low focal length of camera optics and lenses, blurring, improper focusing, atmospheric scattering, target motion, motion between the camera sensor and the scene or subject, and insufficient sampling. The objective of this study was to examine the effect of single- and multi-frame super resolutions on satellite ima- ges and assess the classified results on parameters of image quality. The underlying idea was to present a comparative view on how classification accuracy can be improved when applied on super-resolved satellite imagery. & Anil Kumar anil@iirs.gov.in Shubham Rana shubhamrana7889@gmail.com 1 Indian Institute of Remote Sensing, IIRS/ISRO, 4-Kalidas Road, Dehradun 248 001, India 123 Journal of the Indian Society of Remote Sensing https://doi.org/10.1007/s12524-019-01026-1