Week 5 - Kinematic motion Analysis 2

Question 1

• Briefly explain the difference between Optoelectric, Automatic and Real-time automatic motion analysis systems:

Answer 1

• optoelectric systems have active (infrared signals) and passive (reflective signals) markers
• Automatic has auto marker tracking, auto prediction and marker search
• Real-time uses either video or IR-sensitive cameras, can generate capture and reconstructed output in real time (very slight delay)

Question 2

Explain the main differences between 2D and 3D data collection procedures.

Answer 2

• 2D: generally requires the use of only one camera, Data usually represented in x and y direction
• 3D: This requires a minimum of 2 cameras (often more complex movements), data in x, y, z directions, gen-locked for synchronisation (shutter, frames)

Question 3

Outline the advantages and disadvantages of using a 3D data collection system versus a 2D system.

Answer 3

Advantages of 3D over 2D:
- accurate representation of movement in all directions (x,y,z)
- Rotational movements can be analysed
- Can extrapolate data to obtain rotational kinematic and kinetic data
- Allows more complete and accurate analysis and refinement of movement patterns
- Disadvantages of 3D over 2D:
- Relatively expensive
- computer hardware and software is expensive
- complex nature of set-up of data collection and use of software
- Must be in laboratory setting, limited

Question 4

List several points of consideration when performing 2D data collection for each of the following:

Answer 4

Cameras:
- tripod, levelled, no panning
- Positioned far away, zoomed in
- Make image as large as possible
- Ensure plane of motion is perpendicular to the optical axis of camera
- Use sufficient frame rate and shutter speed to capture motion
- Ensure images are recorded!

Calibration
- Scale device to known length, same distance from camera as performer
- Horizontal device may remain throughout performance
- if video, vertically aspect ratio
- Scale device must be large enough
- distortion error
- digitising error

The background area
- Clear, uncluttered, non-reflective
- Avoid filming into the sun
- Ensure adequate light for correct exposure (beware of glare)

The subjects
- Written informed consent
- Form fitting clothing
- Land marks should contrast
- Allow performance to be unobstructed

Question 5

Explain what an analogue to digital converter does.

Answer 5

VHS cameras
- Conversion of physiological signal (Voltage) to a computer signal (binary digital code)
- An electrical signal fed into an A-D converter
- Signal is sampled (series of snapshots)
- Fed to computer memory
- Analysed by us

Question 6

Explain the “Sampling Theorem” using video collection as an example.

Answer 6

• Digital signals consist of a series of snapshots of a physiological signal taken at regular time intervals
• How often in 1 second these snapshots are taken is called the sampling rate (eg 50Hz)
• “The process signal must be sampled at a frequency at least TWICE as high as the highest frequency present in the signal itself”
• The human eye averages or smoothes out the jumping movement
• Too low: analysis errors & false frequencies
• Too high: more expensive equipment to sample at high frequency as well as analyse

Question 7

What is digitising?

Answer 7

Process of converting film/video images into computer images with known co-ordinate geometry

Question 8

List and explain several sources of error when digitising and how these errors might be minimised.

Answer 8

• Grain / pixels
• Marker shift
• Segmental data assumptions
• Motion outside plane perpendicular to camera axis (2D)
• Motion outside calibration volume
• Data filtering/smoothing technique
• Calculation of derived quantities
• Timing calibration

Question 9

What minimum number of reference control points is required for the solution of Direct Linear Transformation?

Answer 9

A min number of six control points on a reference structure with known X, Y and Z co-ordinates is needed for the solution of the DLT.

Question 10

Explain the statement: Accurate co-ordinate reconstruction can only be guaranteed within the calibration volume.

Answer 10

• Accurate co-ordination reconstruction can only be guaranteed when there is an object with a known reference point

Question 11

• Explain the problems associated with over and under smoothing

Answer 11

• If the choice of cut off frequency is too low or the data is over smoothed then valuable data will be lost
• If the cut off frequency is too high or the data is not smoothed enough then unwanted noise will remain.
• The degree of smoothing from digitised data is often subjective – visual observation decision

Question 12

There are several forms of data smoothing and filtering that can be applied to kinematic motion raw data. Be able to name and explain several of these processes.

Answer 12

• Finite difference technique - takes weighted average but not usually good enough if you want acceleration
• Digital low pass filters - Widely used to filter/remove high frequency noise from data but chops out sudden changes
• Fourier series truncation - converts noisy data into frequency domain, presented as amplitudes of frequency, filtered to remove high frequency noise, transformed back into time domain.
• Curve fitting - involves fitting a smooth curve (drawing a line of best fit) to a series of data points (cubic, quintic, polynomial)

Question 13

Curve fitting smoothing techniques (3):

Answer 13

Cubic spline
- Uses 3 data points
- Forces data to meet at end points
Quintic spline
- Uses 5 data points
- Forces data to meet at end points
Polynomial spline
- 1 equation for whole data set
- First order = straight line
- Better for predictable motion