Autoradiography is a common method in biomedical research for detecting and measuring biodistributions of labelled biomolecules within a specimen. The conventional method is based on using film or film-emulsions for the image acquisition. Although film autoradiography is still in widespread use, there are several disadvantages such as long exposure times, lack of sensitivity, non-linear response of the film and limited dynamic range that encouraged the development of digital autoradiographic systems. Most of the current digital imaging systems have demonstrated excellent performance as far as the above parameters are concerned but still cannot match the image resolution performance exhibited by film or film-emulsion. This thesis is focused on developing image processing methods for improving the quality of digital autoradiography images corrupted with noise and blur obtained by a hybrid CCD autoradiography system at room temperature. Initially, a novel fixed pattern noise method was developed which takes into account the non-ergodic nature of the dark current noise and its dependence on ambient temperature. Empirical formulae were also deduced as a further improvement of the above method for adapting the parameters of the noise distribution for ambient temperature shifts. Image restoration approaches were developed using simulated annealing as a global optimisation technique appropriate for removing the noise and blur from high particle flux samples. The performance of the proposed methods for low flux distributed sources (microscales and mouse brain sections) labelled with high energy beta emmiters has also been demonstrated at different temperatures and integration times and compared with images acquired by the conventional film-based method. Key words: Digital autoradiography, image restoration, simulated annealing, fixed pattern noise removal.