Confocal Microscopy Image Restoration And Regularization Methods

As modern microscopic imaging equipment, confocal laser scanning microscope,with its high depth resolution,is a powerful tool for biomedical and life sciences and other fields. However, in confocal laser scanning microscope imaging acquisition process, due to diffraction and imaging system noise and other factors, the quality of the image has been inevitably degraded. Post-processing image restoration technology, as means to eliminate noise and image blur, are an effective way of restoring sample fluorescent objects. Regularization method introduced certain constraints to convert ill-posed image into well-posed image. It is an effective method of solving the ill-posed image restoration problem.Therefore,in this paper, we present in-depth study of confocal laser scanning fluorescence microscopy image restoration method The main work is listed as follows:(1)The traditional Wiener filtering image restoration algorithm is studied, and proposed a new way of using adaptive value instead of fixed value of the original method. Simulation results show that, compared to Wiener filter algorithm, this method restored image detail more clearly, the overall visual effect is better(2)By adding the regularization parameter in traditional NAS-RIF image restoration algorithm, we improved NAS-RIF algorithm. Experiments show that, the improved algorithm increased clarity of image detail ,it can achieve better recovery of microscopic image(3)Studied the traditional R-L image restoration algorithm, using maximum entropy regularization method for improving the R-L algorithm. Experiments show that the improved algorithm basically eliminated blur in original image and restore more detail of target, but there have been many false impression of tiny structures.(4)Studied the wavelet image restoration algorithm, adopting frequency-domain regularization method for improving the wavelet image restoration algorithm. Experiments show that the improved algorithm of image restoration is better in visual indicators and PSNR.Experiments of simulated images and real confocal laser scanning microscope image. results show that the improved algorithm for image restoration was better than the original algorithm in peak signal to noise ratio and optimum visual effect. It created the conditions for analysis of biological microscopic images.