2.6. Adaptive thresholding algorithm
The major purpose of adaptive thresholding is to give each
pixel a suitable threshold value that is dependent on the distribution of grey levels of the neighbourhood pixels. To achieve
this, we create a map of threshold values which has the same
size with the original image during adaptive thresholding process.
The map stores the threshold values corresponding to
each pixel in the original image and is used to create a binary
image for later processing. To calculate the threshold values
of the map, we adapted the approach developed by Chow
and Kaneko (1972) for outlining boundaries of the heart ventricle
from X-ray image. The X-ray image is initially divided
into many M×M (M= 32 in most cases) operational regions as
shown in Fig. 7a. A threshold value is then determined from
the histogram of this M×M sub-image using an automatic
thresholding algorithm. The algorithm is an unsupervised
thresholding method by iteratively choosing the threshold
value Ts with the following equation (Gonzales and Woods,
2002)