I try to understand the concept behind Wii device. It has 4 camera with capture user's movement, I wonder how does these cameras work? Any reference, article, or papers would be great source for me to start. A mathematical explanation would be greatly appreciated.
From the description you gave it sounded more like the Kinect you were describing (that doesn't have 4 cameras either though, unless for the Wii you're referring to a sensor in each of the 4 controllers). Here's a good wiki explanation of how the controller works.
TL;DR; - An optical sensor can see the LED's in the sensor bar and based on the known distance between the LED's at either end it can work out distance, plus orientation and relative position.
Wii uses 3 mechanisms to track its controller:
Basic implementation was a simple optical calculation of remote's orientation using infrared-sensitive "camera" on top of a controller.
With MotionPlus gyroscope and accelerometer work together (similarly to a smartphone) to detect remote's movement and rotations. It's a much more precise measurement and you don't have to point the controller towards the sensor bar.
The only problem is, it's a relative, not absolute changes we detect, so after some time measurement errors start to add up.
So once in a while infrared camera kicks in and recalibrates the system to fit the LEDs on sensor bar.