Combined interpolation—restoration of Landsat images through FIR filter design techniques L. M. G. FONSECA, G. S. S. D. PRASAD and N. D. A. MASCARENHAS Instituto Nacional de Pesquisas Espaciais, Avenue dos Astronautas 1758, 12201 São José dos Campos, S. P., Brasil (Received 12 July 1989; in final form 9 October 1992) Abstract. In digital image processing for remote sensing there is often a need to interpolate an image. Examples occur in scale magnification, image registration, geometric correction, etc. On the other hand, this image can be subject to several sources of degradation and it would be interesting to compensate also for this degradation in the interpolation process. Therefore, this article addresses the problem of combining interpolation and restoration in a single operation, thereby reducing the computational effort. This is done by means of two-dimensional, separable, Finite Impulse Response (FIR) filters. The ideal low pass FIR filter for interpolation is modified to account for the restoration process. The Modified Inverse Filter (MIF) and the Wiener Filter (WF) are used for this purpose. The proposed methods are applied to the interpolation-restoration of Landsat-5 Thematic Mapper data. The later process takes into account the degradation due to optics, detector and electronic filtering. A comparison with the Parametric Cubic Convolution (PCC) technique is made. The experimental results consist of interpolation-restoration processes of Landsat-5 Thematic Mapper images from 30 m to 15 m (scale magnification) but they could also be generalized to include deblurring on more general interpolation problems, like geometric correction 1. Introduction The resolution of images obtained by satellite sensors is degraded by sources, such as: optical diffraction, detector size and electronic filtering. As a consequence, the effective resolution is, in general, worse than the nominal resolution, that corresponds to the detector projection on the ground and does not take into consideration the sensor imperfections. Through resoration techniques, it is possible to improve image resolution up to a certain levei. This paper explores the idea of combining the restoration process with an interpolation process to generate images with a better resolution over a finer grid than the original sampling grid. Related works in this area include those of Seto et al. (1990), Malaret (1985), Kalman (1984), Wilson (1979), Dye (1975). The combined interpolation-restoration process is performed by means of two- dimensional, separable, Finite Impulse Response (FIR) filters. The ideal low pass FIR filter for interpolation (Crochiere and Rabiner 1983) is modified to account for the restoration process. The proposed method is applied to the interpolation- restoration of Landsat-5 Thematic Mapper (TM) data. 2. The problem of image restoration The image restoration problem attempts to recover an image that has been degraded by the limited resolution of the sensor as well as by the presence of noise. 0143-1161/93 $10.00 (2) 1993 Taylor & Francis Ltd