Close

Kernel Subtraction for Swipe Detection

A project log for Xtencion

An intelligent wrist wear that has unparalleled functionality, improves user's productivity and provides better control over smartwatches.

Chirag Mahaveer ParmarChirag Mahaveer Parmar 08/07/2018 at 14:390 Comments

A new arduino source code file named Xtencion_KernelSubtraction has been added which implements the kernel subtraction method to detect swipes. So we define a array of distances that best represents a swipe, kernel. The kernel array is of size equal to sample_size. Incoming distance readings are stored in a buffer of size equal to sample_size. The kernel is subtracted from the buffer and the difference is stored in another array. Ideally, the difference array should be an array of zeroes. Although, in real world humans may not be able to reproduce the same swipe repeatedly. Hence, we compute the mean and standard deviation of the difference array. Whenever the mean and standard deviation is inside defined limits a swipe is recognised.

Discussions