Continuous spatial domain image identification and restoration with multichannel applications

This research deals with the problem of identification and restoration of images. The field of image identification involves estimating the properties of an imperfect imaging system from an observed image prior to the restoration process, which is concerned mainly with the recovery or the original image from the corrupted one, given the properties of the imaging system. Thus, the two problems are related to each other in the sense that good restoration results depend on how accurate the identified parameters are to the actual situation. The purpose of this research is to investigate some novel identification techniques and their implementations in monochrome and multichannel image processing.; Using the maximum likelihood estimation (ML) approach, the image is represented as an autoregressive (AR) model and blur is described as a continuous spatial domain model. Such formulation overcomes some major limitations encountered in other ML methods. It is shown that blur extent can be optimally identified from noisy images that are degraded by uniform linear motion and out-of-focus blurs. Also, it is shown that angular linear motion can be recovered by identification in two orthogonal directions. The overall performance of this method in conjunction with restoration and noise sensitivity demonstrates its success.; Extending this approach to multichannel case, the image and blur are modeled to include cross-spectral and spatial components. Such components are inherent to multichannel imaging systems and degrade the image further. It is shown that by evaluating and incorporating those components in the identification and restoration techniques, the overall performance is improved significantly. Application of this method to color images and comparison to the independent restoration approach and other existing techniques are also investigated. The novelty of this approach in identifying the blur of multichannel images is a major contribution in producing visually acceptable results and is a significant step for higher processing levels.