Adaptive thresholding algorithms for images with varying background light
Teck Chew Ng · DR-NTU (Nanyang Technological University) · 1997
The environmental lighting control has always been a major problem in applications involving visual inspection, as the separation of objects from background, in conditions of poor and non uniform illumination, is difficult to achieve. Traditional global thresholding techniques are unable to achieve this task as these techniques usually assume that the image pixels are bi-modal in nature which are always not the case in the real-life vision applications. In such cases, one has to rely on adaptive thresholding methods that gather more illumination information from the given images to help making decision on the object and background segmentation. The process of adaptive thresholding is done automatically without any human intervention.