Resampling considerations for registering remotely sensed images

Geodesy and Cartography

2013

Abstract. Image registration is a key component of spatial analyses that involve different data sets of the same area. Automatic approaches in this domain have witnessed the application of several intelligent methodologies over the past decade; however accuracy of these approaches have been limited due to the inability to properly model shape as well as contextual information. In this paper, we investigate the possibility of an evolutionary computing based framework towards automatic image registration. Cellular Neural Network has been found to be effective in improving feature matching as well as resampling stages of registration, and complexity of the approach has been considerably reduced using corset optimization. CNN-prolog based approach has been adopted to dynamically use spectral and spatial information for representing contextual knowledge. The salient features of this work are feature point optimisation, adaptive resampling and intelligent object modelling. Investigations over various satellite images revealed that considerable success has been achieved with the procedure. Methodology also illustrated to be effective in providing intelligent interpretation and adaptive resampling.

Section: Adaptive Resamplingmentioning

confidence: 99%

A CNN Based Hybrid Approach Towards Automatic Image Registration

Geodesy and Cartography

2013

“…NN selects the nearest pixel value where as Bilinear (BL) uses neighboring two points to compute required value. Cubic Convolution (CC) adopts a 4-point kernel based on cubic splines (Camann et al, 2010). Choice of resampling kernels depends on the intended use of the data.…”

Section: Resamplingmentioning

confidence: 99%

A CNN based Hybrid approach towards automatic image registration

Geodesy and Cartography

Katiyar²

2013

Image registration is a key component of various image processing operations which involve the analysis of different image data sets. Automatic image registration domains have witnessed the application of many intelligent methodologies over the past decade; however inability to properly model object shape as well as contextual information had limited the attainable accuracy. In this paper, we propose a framework for accurate feature shape modeling and adaptive resampling using advanced techniques such as Vector Machines, Cellular Neural Network (CNN), SIFT, coreset, and Cellular Automata. CNN has found to be effective in improving feature matching as well as resampling stages of registration and complexity of the approach has been considerably reduced using corset optimization The salient features of this work are cellular neural network approach based SIFT feature point optimisation, adaptive resampling and intelligent object modelling. Developed methodology has been compared with contemporary methods using different statistical measures. Investigations over various satellite images revealed that considerable success was achieved with the approach. System has dynamically used spectral and spatial information for representing contextual knowledge using CNN-prolog approach. Methodology also illustrated to be effective in providing intelligent interpretation and adaptive resampling.

“…Literature reveals a great deal of image resampling techniques (Camann, Thomas, and Ellis 2010;Inglada et al 2007;Wyawahare, Pradeep, and Hemant 2009); however, selection of an optimal method is a difficult task due to the irregular sampling of satellite images and infinite extent of sinc function. The arrival of new generation of satellite sensors with improved spatial and radiometric resolution has led to greater demands on the resampling algorithms.…”

Section: Introductionmentioning

confidence: 99%

“…Bilinear (BL) is a simple twopoint linear interpolator which uses the two neighbouring points to produce a smoothing effect to the image; however, is not used in most applications as it alters the actual image radiance (Australian Geo-Portal 2012). Cubic convolution (CC) is a four-point kernel based on cubic splines and has been used for remote-sensing image analysis as it provides a reasonable compromise between accuracy and speed (Camann, Thomas, and Ellis 2010;Inglada et al 2007). The CC kernel has a slight edge-enhancing effect on the images and the interpolation errors of the CC kernel are significantly worse than a 16-point kernel.…”

Section: Introductionmentioning

confidence: 99%

Intelligent adaptive resampling technique for the processing of remotely sensed imagery

Katiyar

2014

Annals of GIS

Resampling techniques are being widely used at different stages of satellite image processing. The existing methodologies cannot perfectly recover features from a completely undersampled image and hence an intelligent adaptive resampling methodology is required. We address these issues and adopt an error metric from the available literature to define interpolation quality. We also propose a new resampling scheme that adapts itself with regard to the pixel and texture variation in the image. The proposed cellular neural network (CNN)-based hybrid method has been found to outperform existing methods as it adapts itself with reference to the change in sub pixel variation.