OpenVX provides abstractions (via APIs) for vision kernels. The latest reference guide 1.2 lists current defined functions.
One step in my application is an element-wise image division. I am looking for function such as out(x,y) = in1(x,y) / in2(x,y). It seems that OpenVX does not define this function API. Another case is pixel multiplication with constant, such as out(x,y) = in(x,y) * c. Luckily there is Pixel-wise Multiplication out(x,y) = in1(x,y) * in2(x,y) * scale that allows me to work around my implementation. But that always require two images as input.
I know another option is to build my custom node. But I am still wondering why such functions are not available in the specification, while other similar ones such as addition and subtraction are.