Deconvolution Noise Reduction Algorithm Based On Evaluation Of Transfer Function Used For Confocal Microscopical Image

Confocal microscopy imaging techniques are widely used in the fields of precision measurement and biomedical sciences, due to its characteristics of high-contrast lateral resolution,nanoscale axial resolution and optical tomography.The obtained image in confocal microscopy is the degraded image since there are many degradation factors, such as the noise of light source, the aberration of the optical system, the noise of the sensor, the sensor resolution,accurate autofocus and other factors. Currently, a lot of researches have been done on the recovery of the degraded confocal microscopy images. One effective way among these researches is the use of deconvolution algorithm for image noise reduction, which could improve the signal to noise ratio so as to improve the image quality.Firstly, the mathematical model of image degradation mechanisms and restoration of confocal microscopy images are studied. Confocal microscopy image degradation mechanisms as well as the point spread function(PSF)estimate which are commonly used in imaging systems are summarized. Then several common image restoration methods are discussed and analyzed. Their principles and characteristics are also described.Secondly, the restoration quality of confocal microscopy images after they are processed by the deconvolution noise reduction algorithms is studied. A modulation transfer function is used to evaluate the restoration quality of the images based on the summarization of the currently used image quality evaluation methods. In this new method, the curve of the modulation transfer function of the imaging system is depicted through the contrast of the obtained image. The areas, which are surrounded by the transfer function curve and the axes before and after the processing of the deconvolution noise reduction algorithms, are compared to evaluate the effect of these deconvolution algorithms.Then, the deconvolution noise reduction algorithms for the restoration of confocal microscopy images are further studied. Detailed studies are done on the principle of the constrained least squares method, Lucy-Richardson algorithm,blind deconvolution algorithm and the iterative blind deconvolution algorithm.These algorithms are applied to process the confocal microscopy images and significant improvement of the image quality can be found.Finally, the typical line pairs of a discrimination plate with different spatial frequency are imaged using the confocal microscopy system. Then the four different deconvolution noise reduction algorithms which are described above are applied to process the images of these line pairs. The performance of these algorithms are evaluated based on the areas surrounded by the curve of the modulation transfer function and the axes before and after the image processing.