You have to select them based on the image content and your own intent. Refer to the documentation:
ws is the filter window size (pixels). It's just a scalar. The window size is square.
C is an offset parameter used to avoid thresholding noise in uniform regions where there is no object content within the filter window. It's a scalar in the range [0 1]. It's generally going to be a small value. Note that the examples in the documentation seem to describe numbers on a uint8 [0 255] scale instead of [0 1], as the given implementation uses.
