Binary Machine Vision
For object recognition and detection, the input image can be simplified by generating an output image whose pixels:
Tend to have high values if they are part of an objects of interest (“1” in binary image); and low values if they are not (“0” in binary image).
PS: it can also be “0” for objects and “1” for background.
Thresholding - Labelling
To identify an object, the easiest way is to perform a Thresholding-Labelling operation to produce a binary image.
1.Determine a threshold value.
2.Those pixels having grey value higher than the Threshold value are given the value 1. \Those pixels having grey value lower than the Threshold value are given the value 0.
The generation and analysis of such image are called Binary Machine Vision
Mathematical Definition of Thresholding:
#全局、局部、动态
The determination of a threshold value becomes difficult:
1. When the distributions for the bright and dark pixels become more and more overlapped, the valley between the two histogram modes becomes more difficult, because the valley begins to disappear as the two distributions begin to merge together.
2. Furthermore, when there is substantial overlap, the choice of threshold as the valley point is less optimal in the sense of minimizing classification error.
Th