Graphics Reference
In-Depth Information
Fig. 8.1
A sample tic-tac-toe
game in which X wins
Tic-tac-toe has been the topic of choice for many interesting projects in the past.
In 1952, it became one of the first video games—the EDSAC computer could play
it flawlessly against a human opponent. In 1975, MIT students created a tic-tac-toe
computer [
2
] almost entirely out of Tinkertoy pieces (a type of wooden construction
set for children) that could play a perfect tic-tac-toe game. As far as we know, there
has been no prior work to combine vision and AI for a game of tic-tac-toe.
The objective of this project was to create an Android app that plays tic-tac-toe
against an opponent or against itself. The app should be able to do the following:
“See” the tic-tac-toe board hand-drawn on a sheet of paper
“Understand” the visual input
“Think” of the most optimal next move
Report the result to the user
The human user then writes the move on the game paper and the game goes on.
8.2 Vision Algorithms with a Mobile Twist
In this section, we explain how our choice of vision algorithms for the Tic-Tac-
Tandroid app was influenced by mobile environment constraints.
In many ways, the mobile environment is more restrictive for vision applications
than the embedded vision in general. This is because in embedded vision, we gen-
erally have at least some control over which processor is used for vision, how much
memory, what kinds of optics are designed in, etc. In mobile environment, we are
constrained by what processor is the common denominator for all supported plat-
forms, such as smart phones and tablets, and have to rely on operating systems to
assignmemory to our app. Considering howdata-intensive vision is, these can be seri-
ous impediments in achieving real-time operation. There are not many high-profile
mobile vision apps around, so it may be worth mentioning a few:
Sudoku Grab [
3
], which reads a Sudoku puzzle before solving it.
EyesOnRoad [
4
], an app that can assist drivers and improve road safety.
Many developments around the Google Project Glass [
5
], a platform that opens
up many possibilities for using vision for augmented reality and human-computer
interaction.
