image processing (mainly opencv) indexing conventions -

i trying understand indexing conventions of image processing type functions, in opencv.

so indexing pixel [row, column] i.e. y,x, right? think pretty common in image processing.

but when picking points rectangle in opencv requires [col, row] i.e. x,y. , true function wants cvpoint

so if writing function say, takes sample coordinate in image, should [col, row] or [row, col]?

internally image almost stored in rows - index naturally [row,column]

but in maths x,y more common function asks individual pixel coordinate better f(x,y)

opencv confuses matters creating images call specifies image (rows,cols) ie. image(height,width) nam image cols-rows ie 640x480 or 1920x1080