Detecting Common Gestures

Choosing a gesture detection approach depends on the needs of your project. If the project uses only a few simple gestures, then either the algorithmic or neural network approach is recommended. For all other types of projects, it may be in your best interest to invest the time to build a reusable gesture detection engine or to use one of the few available online. In the sections that follow, we describe several common gestures and demonstrate how to implement them using the algorithmic approach. The other two methods of gesture detection are beyond the scope of this book due to their complex and advanced nature.

Regardless of the system chosen to detect gestures, each must account for variations in the performance of a gesture. The system must be flexible and allow for multiple ranges of motion for the same gesture. Rarely does a person perform the same gesture exactly the same way every time, much less the same as other users. For example, make a circular gesture with your left hand. Now repeat that gesture ten times. Was the radius of the circle the same each time? Did the circle start and end at exactly the same point in space? Did you complete the circle in the same duration each time? Perform the same experiment with your right hand and compare the results. Now grab friends and family and observe them making circles. Stand in front of a mirror and watch yourself gesture. Use a video recorder. The trick is to observe as many people as possible performing a gesture and attempt to normalize the movement. A good routine for gesture dete....