Dear Biomch-L Readers,

I am a post graduate student studying algorithms for 3-D movement analysis systems. I would like to know how one calculates and expresses the spatial resolution of the systems. Here are few examples.

For instance in H.Hatze (1988) HIGH-PRECISION THREE-DIMENSIONAL PHOTOGRAMMETRIC CALIBRATION AND OBJECT SPACE RECONSTRUCTION USING A MODIFIED DLT-APPROACH. J. Biomechanics 21, 533-538 is written :

"The highest reconstruction accuracy of 0.733mm rms mean error was obtained...

..This correspondes (1.How is that correspondence established?) to a spatial resolution of about one part in 2860 or 0.035% overall accuracy."

On the other hand in the paper S. Fioretti et al (1985) STEREOMETRY IN VERY CLOSE-RANGE STEREOPHOTOGRAMMETRY WITH NON-METRIC CAMERAS FOR HUMAN MOVEMENT ANALYSIS. J. Biomechanics, 18. 831-842 is said:

"The attainable accuracy is high, better than 0.1% of the observation distance".

2. How is the observation distance now involved since I can change field of view by changing cameras zoom and maintaing same distance? Would not be field of view more appropriate?

I have found in Andriacchi T.P. et al (1979) THREE-DIMENSIONAL DATA PROCESSING IN HUMAN MOTION ANALYSIS, J. Biomechanocal Engineering, 101. 279-283: "A resolution of one part in 500 was achived over a viewing field of 2.4m". It is stil not clear how they came up with that number?

3. It seems to me that all authors calculate rms values in x,y and z directions, take into account field of view and/or camera (which camera) distance and give the spatial resolution? But how it is done?

4. Some authors are also mentioning confidence intervals. Is there any standard (standard test or so) for expressing systems spatial accuracy therefore when I am bying one I will know where I am at?

Thank you all in advance. I will be glad to publish a summary of responses later on.

Tomislav Pribanic

Department for Electronic Systems and Information Processing

Faculty of Electrical Engineering and Computing

3 Unska, 10000 Zagreb, Croatia

E-mail : tomislav.pribanic@zesoi.fer.hr

Dear Biomch-L Readers,

I am a post graduate student

studying algorithms for 3-D movement analysis systems. I would like to know how

one calculates and expresses the spatial resolution of the systems. Here are few

examples.

For instance in H.Hatze

(1988) HIGH-PRECISION THREE-DIMENSIONAL PHOTOGRAMMETRIC CALIBRATION AND OBJECT

SPACE RECONSTRUCTION USING A MODIFIED DLT-APPROACH. J. Biomechanics 21,

533-538 is written :

"The highest reconstruction accuracy of

0.733mm rms mean error was obtained...

...This correspondes (1.How is that

correspondence established?) to a spatial resolution of about one part in 2860

or 0.035% overall accuracy."

On the

other hand in the paper S. Fioretti et al (1985) STEREOMETRY IN VERY CLOSE-RANGE

STEREOPHOTOGRAMMETRY WITH NON-METRIC CAMERAS FOR HUMAN MOVEMENT ANALYSIS. J.

Biomechanics, 18. 831-842 is said:

"The attainable accuracy is high, better than 0.1% of the

observation distance".

2. How is the observation distance now

involved since I can change field of view by changing cameras zoom and maintaing

same distance? Would not be field of view more appropriate?

I have found in Andriacchi T.P. et al (1979)

THREE-DIMENSIONAL DATA PROCESSING IN HUMAN MOTION ANALYSIS, J. Biomechanocal

Engineering, 101. 279-283: "A resolution of one part in 500 was achived

over a viewing field of 2.4m". It is stil not clear how they came up with

that number?

3. It

seems to me that all authors calculate rms values in x,y and z directions, take

into account field of view and/or camera (which camera) distance and give the

spatial resolution? But how it is done?

4. Some authors are also mentioning

confidence intervals. Is there any standard (standard test or so) for expressing

systems spatial accuracy therefore when I am bying one I will know where I am

at?

Thank you all in advance. I will be glad to

publish a summary of responses later on.

Tomislav PribanicDepartment for Electronic

Systems and Information ProcessingFaculty of Electrical Engineering and

Computing3 Unska, 10000 Zagreb, CroatiaE-mail : tomislav.pribanic@zesoi.fer.hr

I am a post graduate student studying algorithms for 3-D movement analysis systems. I would like to know how one calculates and expresses the spatial resolution of the systems. Here are few examples.

For instance in H.Hatze (1988) HIGH-PRECISION THREE-DIMENSIONAL PHOTOGRAMMETRIC CALIBRATION AND OBJECT SPACE RECONSTRUCTION USING A MODIFIED DLT-APPROACH. J. Biomechanics 21, 533-538 is written :

"The highest reconstruction accuracy of 0.733mm rms mean error was obtained...

..This correspondes (1.How is that correspondence established?) to a spatial resolution of about one part in 2860 or 0.035% overall accuracy."

On the other hand in the paper S. Fioretti et al (1985) STEREOMETRY IN VERY CLOSE-RANGE STEREOPHOTOGRAMMETRY WITH NON-METRIC CAMERAS FOR HUMAN MOVEMENT ANALYSIS. J. Biomechanics, 18. 831-842 is said:

"The attainable accuracy is high, better than 0.1% of the observation distance".

2. How is the observation distance now involved since I can change field of view by changing cameras zoom and maintaing same distance? Would not be field of view more appropriate?

I have found in Andriacchi T.P. et al (1979) THREE-DIMENSIONAL DATA PROCESSING IN HUMAN MOTION ANALYSIS, J. Biomechanocal Engineering, 101. 279-283: "A resolution of one part in 500 was achived over a viewing field of 2.4m". It is stil not clear how they came up with that number?

3. It seems to me that all authors calculate rms values in x,y and z directions, take into account field of view and/or camera (which camera) distance and give the spatial resolution? But how it is done?

4. Some authors are also mentioning confidence intervals. Is there any standard (standard test or so) for expressing systems spatial accuracy therefore when I am bying one I will know where I am at?

Thank you all in advance. I will be glad to publish a summary of responses later on.

Tomislav Pribanic

Department for Electronic Systems and Information Processing

Faculty of Electrical Engineering and Computing

3 Unska, 10000 Zagreb, Croatia

E-mail : tomislav.pribanic@zesoi.fer.hr

Dear Biomch-L Readers,

I am a post graduate student

studying algorithms for 3-D movement analysis systems. I would like to know how

one calculates and expresses the spatial resolution of the systems. Here are few

examples.

For instance in H.Hatze

(1988) HIGH-PRECISION THREE-DIMENSIONAL PHOTOGRAMMETRIC CALIBRATION AND OBJECT

SPACE RECONSTRUCTION USING A MODIFIED DLT-APPROACH. J. Biomechanics 21,

533-538 is written :

"The highest reconstruction accuracy of

0.733mm rms mean error was obtained...

...This correspondes (1.How is that

correspondence established?) to a spatial resolution of about one part in 2860

or 0.035% overall accuracy."

On the

other hand in the paper S. Fioretti et al (1985) STEREOMETRY IN VERY CLOSE-RANGE

STEREOPHOTOGRAMMETRY WITH NON-METRIC CAMERAS FOR HUMAN MOVEMENT ANALYSIS. J.

Biomechanics, 18. 831-842 is said:

"The attainable accuracy is high, better than 0.1% of the

observation distance".

2. How is the observation distance now

involved since I can change field of view by changing cameras zoom and maintaing

same distance? Would not be field of view more appropriate?

I have found in Andriacchi T.P. et al (1979)

THREE-DIMENSIONAL DATA PROCESSING IN HUMAN MOTION ANALYSIS, J. Biomechanocal

Engineering, 101. 279-283: "A resolution of one part in 500 was achived

over a viewing field of 2.4m". It is stil not clear how they came up with

that number?

3. It

seems to me that all authors calculate rms values in x,y and z directions, take

into account field of view and/or camera (which camera) distance and give the

spatial resolution? But how it is done?

4. Some authors are also mentioning

confidence intervals. Is there any standard (standard test or so) for expressing

systems spatial accuracy therefore when I am bying one I will know where I am

at?

Thank you all in advance. I will be glad to

publish a summary of responses later on.

Tomislav PribanicDepartment for Electronic

Systems and Information ProcessingFaculty of Electrical Engineering and

Computing3 Unska, 10000 Zagreb, CroatiaE-mail : tomislav.pribanic@zesoi.fer.hr