Code Sharing is Associated with Research Impact in Image Processing

In computational sciences such as image processing, the publication itself is often not enough to allow other researchers to verify the results by repeating the described experiments. In many cases, supplementary material such as source code and measurement data are required, or can at least be very helpful. Still, only approximately 10% of recently published papers in image processing have code available online. One of the arguments for not making code available is the extra time required to prepare the material. In this paper, we claim that this additional time may be well spent, as the availability of code for a publication is associated with an increase in the expected number of citations. We show this with exploratory analyses of the relationship between code availability and the number of citations for image processing papers.