Otsu's algorithm
In image processing and segmentation, the Otsu's algorithm is very important to select a suitable
threshold level and extract the objects from their background. This algorithm allows to find a
threshold value as a function of time, so that it minimizes the variance of weights between classes,
i.e., the optimal threshold value is automatically calculated based on the input image.
In this way the rest of the scenario is limited by means of functions that allow to fill the black
points that are larger and to eliminate the white points that are smaller obtaining a new mask, to
then use the operation of intersection of sets and measure a physical variable with very high
precision. Also based on the compression of horizontal, vertical and diagonal segments the value
of the variable is obtained by making a comparison between the different classes and storing each
vector of points in an output vector containing information about the topology of the image.
Finally, the optimal threshold is selected by maximizing the separability measure of the resulting
classes in gray levels, based on a simple procedure that covers a wide range of unsupervised
decisions.
The applications of Otsu's algorithm are not restricted only to thresholding images in the gray
scale, but can also help in the classification of objects with some discriminative feature and provide
a solution to several practical problems.
Reference
Otsu, N. (1979). A Threshold Selection Method from Gray-Level Histograms. IEEE Transactions
on Systems, Man, and Cybernetics, 9(1), 62–66. doi:10.1109/tsmc.1979.4310076
In image processing and segmentation, the Otsu's algorithm is very important to select a suitable
threshold level and extract the objects from their background. This algorithm allows to find a
threshold value as a function of time, so that it minimizes the variance of weights between classes,
i.e., the optimal threshold value is automatically calculated based on the input image.
In this way the rest of the scenario is limited by means of functions that allow to fill the black
points that are larger and to eliminate the white points that are smaller obtaining a new mask, to
then use the operation of intersection of sets and measure a physical variable with very high
precision. Also based on the compression of horizontal, vertical and diagonal segments the value
of the variable is obtained by making a comparison between the different classes and storing each
vector of points in an output vector containing information about the topology of the image.
Finally, the optimal threshold is selected by maximizing the separability measure of the resulting
classes in gray levels, based on a simple procedure that covers a wide range of unsupervised
decisions.
The applications of Otsu's algorithm are not restricted only to thresholding images in the gray
scale, but can also help in the classification of objects with some discriminative feature and provide
a solution to several practical problems.
Reference
Otsu, N. (1979). A Threshold Selection Method from Gray-Level Histograms. IEEE Transactions
on Systems, Man, and Cybernetics, 9(1), 62–66. doi:10.1109/tsmc.1979.4310076
