Just an adaptive noise filter for numerical data. It’s a noise reduction test for a project I’m working on irl. Noise variance is currently +-30 units. Value at the top is true variance after filtering.
- note: filter only works for numerical data such as distances, as far as I know. It’s also susceptible to data lag, so is best suited for scenarios with little movement in the data. It does its best to compensate, but it’s not perfect. - Blue line is the true data, and blue dot is the last measured data point (ie. The current position of your mouse). - Red line is the data as if picked up by a sensor, with noise. - white line is the estimate after applying the EMA filter. - Green “reticle” is the output, an averaged and fairly accurate estimate from very noisy data. Should be close to the blue dot, numerically displayed at the top. Most of the time.
