PDF version of this document
Work in the field seems to move in differentiable yet almost identical directions. On the one hand, speed is an issue that might not have a solution and Section 7 elaborates on imaginable hindrances of this kind. Algorithmic trade-offs and different choices of programming language and paradigm, operating system and platform is a matter worth pursuing. Since the process relies on wide global scope, i.e. investigation of various images simultaneously where change in one affects all, heuristics can perhaps be applied to decrease the number of iterations involved42.
Claims of a similar nature can be made on the more intrinsic part of work being reviewed. For a start, values were often tweaked manually and no strong evidence was used to support such arbitrary selections which purportedly followed common sense. Another problem that has been realised is that much of the process comprised the simplistic joining of remotely-related components whose nature is unique and autonomous. This means that components in the system often suffered from the undependable fusion which was in place - something which is a direct consequence of the knowledge that is still lacking in the field. There is much to be learned about how the numerous existing techniques, measurements and component should be merged effectively and by all means conveniently.