What does oplus symbol ⊕ do for 2 images in Convolution Neural Networks

Question

So I'm reading this paper on optical flow prediction from two image frames, and I'm having a difficult time finding what this operator does. This paper, and some other ones uses it on the outputs of convolution neural networks.

http://openaccess.thecvf.com/content_cvpr_2017/papers/Ranjan_Optical_Flow_Estimation_CVPR_2017_paper.pdf

I looked up math symbols and saw that it was an xor operator in logic, but that doesn't make any sense to me in this context.

I would also like to know what to call this symbol for future reference.

Answer 1

The symbol is called "oplus" or "circled plus". In TeX it is written \oplus and looks like this: $$\oplus$$

Depending on context, it could denote

• exlusive or (XOR)
• dilation (morphology)
• direct sum

Here is the disambiguation page.

---

I don't know anything about morphology.

However, given that your question is related to handling images mathematically, I'd guess that dilation is what your $\oplus$ denotes.

Answer 2

So I asked a colleague about the operation and he says the oplus symbol refers to concatenation of the last dimension in the area of computer vision.

So if image1 had shape (w,h,3) and image2 had shape (w,h,3)

image1 $\oplus$ image2 = image3 with shape (w,h,6)

I'd like to confirm whether or not this is true.