A quality consensus strategy for binarization of neuronal images

Abstract

In the field of neural processing, the dendritic spines are likely to be of major importance. The morphology of dendritic spines is very diverse and changes in spine size, as well as their density, are thought to reflect changes in the strength of the synaptic transmission. This paper presents a modified image binarization technique for the neuronal image. Binarization is often recognized to be one of the most important steps in most high-level image analysis systems, particularly for object recognition. It's precise functioning highly determines the performance of the entire system. Although so much research work has been done so far in the field of image binarization still there is a scope for improvement. All of those methods perform well to some extent but in every case, there are a lot of disconnected spines, deformities in dendritic shape and size, and a lot of manual work needs to be done. Image binarization is the process of converting a greyscale image to a black and white image. This image binarization technique is used as a pre-processing step of image analysis in various domains. Thresholding is a method that is used for binarization. In thresholding, an optimal threshold value is chosen and the pixels are classified as foreground or background by comparison with this threshold value. In this paper, we have used local thresholding. Locally adaptive image binarization with a sliding-window threshold can be an effective tool for various image processing tasks. The major focus of this thesis report is to provide a new technique for binarization of medical images using the classical method of Sauvola with some modifications. The experiment deals with the quality consensus that is if one of the decisions for a particular pixel says to be foreground and the other says to be background then foreground is taken of all the variable window sizes and the precision of the neuronal image is evaluated.