Manual diagonal line

This test is available only when device selected = tablet

You can test for linearity and accuracy by manually generating in fluent movement a diagonal line using a ruler. In this test, the orthogonal error, longitudinal error and number of missing samples are charted as well as logged in to the results window & log file.

o Input device settings: Main tool bar > settings > select input device > settings, select the option 'make recording window real size'.

o Enable Logging: To view the numerical results of the test, you need to first clear the existing log file by File > logging > clear log file. Then enable logging by file > logging > content > select all, followed by, file > logging > log actions to file.

Note: If you are repeating the test, and you want to look only at the newest results, do a File > logging > clear log file.

o Select 'Test device for linearity' > 'Manually draw a diagonal line' > GO.

o The dialog window to enter the sampling rate, pen pressure and resolution is displayed. Enter the values as determined from previous tests.

o Perform Recording: Using a ruler, draw a line from the 0,0 point of the recording window to the top right corner of the window. Make sure that the pen tip is kept touching straightly on the tablet without much tilt. Otherwise, this could lead to linearity errors introduced manually. Also, the movement must be made as smoothly as possible, to avoid introduction of errors.

o The raw data plot is displayed.

Raw data charts: They are displayed automatically after the recording, also shows 3 interesting plots with the following parameters plotted against time samples (navigate using 'Next chart' button):

1 a. Orthogonal Error- Horizontal Spacing between successive samples (X coordinate)

1 b. Longitudnal error- Intersample Distance (Y coordinate), which is also the vertical velocity per sample. Go to the chart of x Vs. y > right click > plotting method > points + line. See if the samples are spaced uniformly.

Note approximately the average level of the curve (average vertical velocity). Pen speed (driver properties) should be set such that this average value is higher than 10.

2. Number of missing samples: The number of missing samples from the tablet can be viewed with 1 sample as reference.

The chart values indicate the following:

0 to 0.58 (2 or more samples extra)
0.58 to 0.75 (1 sample extra)
0.75 to 1.5 (Perfect, 1 indicates NO samples missing)
1.5 to 2.5 (One missing sample)
2.5 to 3.5 (2 missing samples)
3.5 and above (3 or more missing samples)

o In all the above 3 charts, the error values (y-axis data) are all zeros, for samples which have an absolute velocity smaller than the average absolute velocity. This is done to ensure that the calculations are performed only for the movement data and not the noise/pen-up movements.

Numerical data/View log file or Results window: The log file and results window now contain all the numerical values from the test/record input device. Log file can be viewed by File > logging > view log file.

o Click here for an example of the resultant log file

The relevant information from the above file have summarized below:

/*** SETTINGS ***/
Resolution = 0.00254
Sampling rate = 75
Minimum Pen Pressure = 1

/*** SAMPLES ***/
Input # of samples = 345
# of samples after trailing penlift removed = 265
#Samples of Pendown = 265 = 3.53 s
First sample coordinates=( 45, 1, 4)
Last sample coordinates=( 8987, 5926, 145)
xlength = 8942 points
ylength = 5925 points
absolute stroke length = 10726.8 points
xlength = 22.71 cm
ylength = 15.05 cm
absolute stroke length = 27.25 cm
Average velocity = 40.6 points/sample = 7.740cm/s.
First 10 consecutive samples above average velocity starts at sample# 13
Last 10 consecutive samples above average velocity starts at sample# 254
Movement range=#of samples above average velocity=240 (Only these will be considered for further testing)

/*** SPATIAL ERROR ***/
RMS sample error = 2.7 points = 0.00686 cm
RMS quantization error = 0.289 points = 0.000733 cm
RMS sample error relative to quantization error = 9.35
RMS sample error without quantization error = 2.68366 points = 0.0068165 cm
Length of fitted line = 10116 points = 25.7 cm; Straightness Error = 0.000267
Average velocity of the fitted stroke = 43.1 points/sample = 7.963cm/s.

/*** TEMPORAL ERROR ***/
#Samples below average velocity of movement=99 (Excluded from missing samples estimation)
Frequency of triple or more points=0
Frequency of double points=0
Frequency of no missing samples=140
Frequency of 1 missing sample=1
Frequency of 2 missing samples=0
Frequency of 3 or more missing samples=0
Total Frequency of missing samples=1
Total Number of missing samples=1
Total Frequency of multiple samples=0
Total Number of multiple samples=0

Results (tablet)

The above test was performed two or more times, and results from one is presented for each tablet in table below:

Tablet make	Wacom	Wacom	Aiptek	GTCO Calcomp
Type	Graphire 3	Intuos 2	8000U	CAD Pro
Size (WxH")	8x6"	8x6"	8x6"	9x6"
RMS sample error (cm)	0.00811	0.00685	0.0136	0.00827
#Samples in movement range	224	133	205	144
Number of indeterminate samples	2	0	0	0
Number of samples missing	0	 0	53	4
Number of multiple samples	1	 0	20	1
Frequency of Missing Samples	No samples missing->221 1 sample missing-> 0 2 samples missing-> 0 3 or more samples missing-> 0	No samples missing->122 1 sample missing-> 0 2 samples missing-> 0 3 or more samples missing-> 0	No samples missing->96 1 sample missing-> 53 2 samples missing-> 0 3 or more samples missing-> 0	No samples missing->136 1 sample missing-> 4 2 samples missing-> 0 3 or more samples missing-> 0
Frequency of Multiple Samples	Doubles->0 Triples->1	Doubles->0   Triples->0	Doubles->9 Triples->47	Doubles->3 Triples->1

 

Real-Life testing

Aim of this test is to replicate the behavior of the tablet from within the program by data generation and simulation (using the device parameters specified and the noise value calculated from previous test)

Generate Data

First, make a note of the user root folder, where all the current user data is stored on your computer, (can be determined from files > users > click user ID > properties > settings)

1. Settings > Test/Record input device > Test device for linearity > Generate a diagonal line

2. To generate 8 circles of 1 cm diameter at 5 Hz, Click the 'settings' button > predefined = custom > fill out the following settings

a. with no noise:

Noise = 0, Device Settings => uncheck the option 'Round raw data', Start time = 0, stroke duration = 0.2, number of strokes = 16, trail time = 0, X-movement pattern (position start = 0.3, size of stroke = 1, start phase = 90, velocity = 0), Y-movement pattern (position start = 0.3, size of stroke = 1, start phase =0)

Click ok > Click 'Go' > Input the actual resolution, sampling rate and pen pressure for the tablet

3. On the test/record window click ok to display the chart. Leave the chart open.

4. Go to the user root folder in windows explorer

5. Navigate to test.hwr > open > save as EEEGGGSSSCCC01.hwr in any local folder (say 'trials')

6. Repeat steps 1. - 5. saving the files as EEEGGGSSSCCC02.hwr, EEEGGGSSSCCC03.hwr, EEEGGGSSSCCC04.hwr EEEGGGSSSCCC05.hwr and EEEGGGSSSCCC06.hwr using each of the following settings and instructions for step 2.

b. With quantization noise and no white noise

Noise = 0, Device Settings => check the option 'Round raw data', Start time = 0, stroke duration = 0.2, number of strokes = 16, trail time = 0, X-movement pattern (position start = 0.3, size of stroke = 1, start phase = 90, velocity = 0), Y-movement pattern (position start = 0.3, size of stroke = 1, start phase =0)

c. With no quantization noise and white noise = RMS noise obtained from previous test

Noise = RMS noise, Device Settings => uncheck the option 'Round raw data', Start time = 0, stroke duration = 0.2, number of strokes = 16, trail time = 0, X-movement pattern (position start = 0.3, size of stroke = 1, start phase = 90, velocity = 0.4), Y-movement pattern (position start = 0.3, size of stroke = 1, start phase =0)

d. With quantization noise and white noise = RMS noise obtained from previous test

Noise = RMS noise, Device Settings => check the option 'Round raw data', Start time = 0, stroke duration = 0.2, number of strokes = 16, trail time = 0, X-movement pattern (position start = 0.3, size of stroke = 1, start phase = 90, velocity = 0.4), Y-movement pattern (position start = 0.3, size of stroke = 1, start phase =0)

e. with no noise and one in 5 samples missing:

Noise = 0, Device Settings => uncheck the option 'Round raw data', Start time = 0, stroke duration = 0.2, number of strokes = 16, trail time = 0, X-movement pattern (position start = 0.3, size of stroke = 1, start phase = 90, velocity = 0), Y-movement pattern (position start = 0.3, size of stroke = 1, start phase =0)

On the test/record window click the button view/edit.

In the test.hwr displayed, remove 1 in every 5 samples (Click here for a procedure to automatically do this. Extract the .zip file contents, read readme file for instructions)

Save the file > close file. Click ok on the test/record window to display the generated data chart.

f. with no noise and random samples missing:

Noise = 0, Device Settings => uncheck the option 'Round raw data', Start time = 0, stroke duration = 0.2, number of strokes = 16, trail time = 0, X-movement pattern (position start = 0.3, size of stroke = 1, start phase = 90, velocity = 0), Y-movement pattern (position start = 0.3, size of stroke = 1, start phase =0)

On the test/record window click the button view/edit.

In the test.hwr displayed, remove random samples (To automatically do this, refer the procedure mentioned in the previous step)

Save the file > close file. Click ok on the test/record window to display the generated data chart.

7. Repeat steps 1. - 6. for 1 Hz frequency movement (that is, to set, stroke duration = 1 sec and x velocity = 0.4 cm/s) for the step 2. settings. Save the files with EEEGGGSSSCCC(07 to 12). hwr respectively.

Process data

1. Create an experiment EEE, with group GGG, subject SSS and condition CCC. Refer to the tutorial to create a minimum-size experiment.

2. Verify in Settings > select input device, the tablet dimensions and mapping. Refer to the section on mapping for mapping options.

3.Experiment > right click > experiment settings > processing > time functions > filter frequency, set to 7 Hz.

4. File > import > data import wizard > data type = Movalyzer, path = select the path to the 'trials' folder with all the trials > select experiment EEE, group GGG, condition CCC, and subject SSS > click finish.

5. All the 12 trials should now appear under the experiment when the left tree is expanded.

6. Right click subject SSS > Reprocess trials.

Analyse data

To summarize, the trials indicate the following:

EEEGGGSSSCCC01 - 5 Hz data with no noise

02 - b. With quantization noise and no white noise

03 - c. With no quantization noise and white noise = RMS noise obtained from previous test

04 - d. With quantization noise and white noise = RMS noise obtained from previous test

05 - e. with no noise and one in 5 samples missing

06 - f. with no noise and random samples missing

Similarly for trials 7-12 with 1 Hz data

Trial > right click > chart processed data > normalized jerk vs. time: This chart shows the effect of the above scenarios on the normalized jerk.

For example, the presence of RMS noise (introduced by the noise from tablet) might increase the jerk to a certain extent. Also the missing samples result in a huge jerk value. Thus, the comparison of the different trials yields to an understanding of the effect of linearity error and missing samples on the data, for the particular tablet under test.

IV. Results

Tablet make	Wacom		Wacom		Aiptek		GTCO Calcomp
Type	Graphire 3		Intuos 2		8000U		CAD Pro
Size (WxH")	6x8"		6x8"		6x8"			-
Average Normalized Jerk per trial	5 Hz	1 Hz	5 Hz	1 Hz	5 Hz	1 Hz	5 Hz	1 Hz
	7.7586	7.9855	7.7586	7.9855	7.7586	7.9855	7.7586	7.9855
	7.9857	10.0705	7.7510	10.0705	7.7510	10.0705	7.7510	10.0705
	7.7941	102.8650	7.8230	132.6889	7.8881	223.5679	7.7950	121.8871
	7.7876	104.6471	7.8203	120.6653	7.9066	208.205	7.7950	133.1956
	7.7506	8.1397	7.7506	8.1397	7.7506	8.1397	8.7541	8.1397
	8.7897	81.8999	8.7897	79.8216	8.7897	81.8999	8.7541	81.8999