Hi Clinton,
On Wed, Mar 12, 2014 at 06:16:21PM -0500, Clinton Sprain wrote:
> input: appletouch: implement sensor data smoothing
>
> Use smoothed version of sensor array data to calculate movement and add weight
> to prior values when calculating average. This gives more granular and more
> predictable movement.
>
> Signed-off-by: Clinton Sprain <clintonsprain@xxxxxxxxx>
> ---
> drivers/input/mouse/appletouch.c | 72 ++++++++++++++++++++++++++++----------
> 1 file changed, 53 insertions(+), 19 deletions(-)
>
> diff --git a/drivers/input/mouse/appletouch.c b/drivers/input/mouse/appletouch.c
> index 2745832..e00f126 100644
> --- a/drivers/input/mouse/appletouch.c
> +++ b/drivers/input/mouse/appletouch.c
> @@ -332,7 +332,11 @@ static void atp_reinit(struct work_struct *work)
> static int atp_calculate_abs(int *xy_sensors, int nb_sensors, int fact,
> int *z, int *fingers)
> {
> - int i;
> + int i, k;
> + int smooth[nb_sensors + 8];
> + int smooth_tmp[nb_sensors + 8];
This unfortunately introduces variable-length arraus on stack which
makes sparse unhappy. Can we add these scratch buffers to the device
structure instead?
> + int scale = 12;
Probably better use a define rather than a variable.
> +
> /* values to calculate mean */
> int pcum = 0, psum = 0;
> int is_increasing = 0;
> @@ -344,9 +348,6 @@ static int atp_calculate_abs(int *xy_sensors, int nb_sensors, int fact,
> if (is_increasing)
> is_increasing = 0;
>
> - continue;
> - }
> -
> /*
> * Makes the finger detection more versatile. For example,
> * two fingers with no gap will be detected. Also, my
> @@ -361,27 +362,60 @@ static int atp_calculate_abs(int *xy_sensors, int nb_sensors, int fact,
> *
> * - Jason Parekh <jasonparekh@xxxxxxxxx>
> */
> - if (i < 1 ||
> +
> + } else if (i < 1 ||
> (!is_increasing && xy_sensors[i - 1] < xy_sensors[i])) {
> (*fingers)++;
> is_increasing = 1;
> } else if (i > 0 && (xy_sensors[i - 1] - xy_sensors[i] > threshold)) {
> is_increasing = 0;
> }
> + }
>
> - /*
> - * Subtracts threshold so a high sensor that just passes the
> - * threshold won't skew the calculated absolute coordinate.
> - * Fixes an issue where slowly moving the mouse would
> - * occasionally jump a number of pixels (slowly moving the
> - * finger makes this issue most apparent.)
> - */
> - pcum += (xy_sensors[i] - threshold) * i;
> - psum += (xy_sensors[i] - threshold);
> + /*
> + * Use a smoothed version of sensor data for movement calculations, to
> + * combat noise without needing to rely so heavily on a threshold.
> + * This improves tracking.
> + *
> + * The smoothed array is bigger than the original so that the smoothing
> + * doesn't result in edge values being truncated.
> + */
> +
> + for (i = 0; i < 4; i++)
> + smooth[i] = 0;
memset might be better?
> + for (i = nb_sensors + 4; i < nb_sensors + 8; i++)
> + smooth[i] = 0;
And here as well.
Thanks.
--
Dmitry
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
