![]() |
ARX
1.0
The next-generation open source augmented reality toolkit.
|
Provides filtering of transformation matrices (pose estimates). More...
#include <ARX/AR/ar.h>Go to the source code of this file.
Macros | |
| #define | AR_FILTER_TRANS_MAT_SAMPLE_RATE_DEFAULT 30.0 |
| Default sample rate (in Hertz). | |
| #define | AR_FILTER_TRANS_MAT_CUTOFF_FREQ_DEFAULT 15.0 |
| Default cutoff frequency (in Hertz). | |
Typedefs | |
| typedef struct _ARFilterTransMatInfo | ARFilterTransMatInfo |
Functions | |
| ARFilterTransMatInfo * | arFilterTransMatInit (const ARdouble sampleRate, const ARdouble cutoffFreq) |
| Initialise a filter for a single transformation matrix (pose estimate). | |
| int | arFilterTransMatSetParams (ARFilterTransMatInfo *ftmi, const ARdouble sampleRate, const ARdouble cutoffFreq) |
| Set the filter parameters. | |
| int | arFilterTransMat (ARFilterTransMatInfo *ftmi, ARdouble m[3][4], const int reset) |
| Filters the supplied pose estimate transformation matrix in-place. | |
| void | arFilterTransMatFinal (ARFilterTransMatInfo *ftmi) |
| Finalise a filter. | |
Provides filtering of transformation matrices (pose estimates).
High frequency noise ("jitter") can be an undesirable property in optical tracking systems like artoolkitX. These functions implement a first-order low-pass filter for both the position and orientation components of the pose estimate transformation matrices.
Usage:
For each pose estimate which is to be filtered, an ARFilterTransMatInfo structure should be allocated by calling arFilterTransMatInit(). Samples are then added to the filter and the filtered value returned by calling arFilterTransMat(). At the end of operations, the filter should be disposed of by calling arFilterTransMatFinal().
| #define AR_FILTER_TRANS_MAT_CUTOFF_FREQ_DEFAULT 15.0 |
Default cutoff frequency (in Hertz).
@defined
This value provides a default value for the cutoffFreq parameter in arFilterTransMatInit().
| #define AR_FILTER_TRANS_MAT_SAMPLE_RATE_DEFAULT 30.0 |
Default sample rate (in Hertz).
@defined
This value provides a default value for the sampleRate parameter in arFilterTransMatInit().
| typedef struct _ARFilterTransMatInfo ARFilterTransMatInfo |
| int arFilterTransMat | ( | ARFilterTransMatInfo * | ftmi, |
| ARdouble | m[3][4], | ||
| const int | reset | ||
| ) |
Filters the supplied pose estimate transformation matrix in-place.
This performs the filter function for a single transformation matrix.
| ftmi | Filter settings to be used with this transformation matrix. |
| m | Transformation matrix representing the current pose estimate. |
| reset | If a discontinuity in transformation matrix pose estimates has occured (e.g. when a marker is first acquired, or is reacquired after a period of not being visible) the filter initial state or "memory" should be set to the same value as the current sample by setting this parameter to 1. If transformation matrix pose estimates have occured continuously, set this parameter to 0. |
| void arFilterTransMatFinal | ( | ARFilterTransMatInfo * | ftmi | ) |
Finalise a filter.
When all filter use has completed, this function should be called to dispose of the filter structure.
| ftmi | Filter settings to dispose of. |
| ARFilterTransMatInfo * arFilterTransMatInit | ( | const ARdouble | sampleRate, |
| const ARdouble | cutoffFreq | ||
| ) |
Initialise a filter for a single transformation matrix (pose estimate).
In order to filter a pose estimate, this function should be called to setup filter data structures.
Each ARFilterTransMatInfo structure should be used only with a single transformation matrix. For example, if you have 5 pose estimates to filter, you would call this function 5 times, once for each pose, and keep the pose data and filter data paired.
The structure produced should be passed to other filter functions dealing with the same transformation matrix.
| sampleRate | This value partially determines the filter properties. It should be as close as possible to the rate at which pose estimates arrive, which is usually the camera frame rate. If you are unsure what value to use, use the constant AR_FILTER_TRANS_MAT_SAMPLE_RATE_DEFAULT. This value may not be 0.0. |
| cutoffFreq | This value partially determines the filter properties. It should be as close as possible to the rate above which you do not wish changes to the incoming transformation matrix pose estimate to be observed. This would usually be the maximum rate at which you imagine to-and-fro rotiational or positional motion of a marker to occur. If you are unsure what value to use, use the constant AR_FILTER_TRANS_MAT_CUTOFF_FREQ_DEFAULT. This value may not be 0.0. |
| int arFilterTransMatSetParams | ( | ARFilterTransMatInfo * | ftmi, |
| const ARdouble | sampleRate, | ||
| const ARdouble | cutoffFreq | ||
| ) |
Set the filter parameters.
This convenience function allows adjustment of the initial filter parameters.
| ftmi | Filter settings of which the parameters should be set. |
| sampleRate | This value partially determines the filter properties. It should be as close as possible to the rate at which pose estimates arrive, which is usually the camera frame rate. If you are unsure what value to use, use the constant AR_FILTER_TRANS_MAT_SAMPLE_RATE_DEFAULT. This value may not be 0.0. |
| cutoffFreq | This value partially determines the filter properties. It should be as close as possible to the rate above which you do not wish changes to the incoming transformation matrix pose estimate to be observed. This would usually be the maximum rate at which you imagine to-and-fro rotiational or positional motion of a marker to occur. If you are unsure what value to use, use the constant AR_FILTER_TRANS_MAT_CUTOFF_FREQ_DEFAULT. This value may not be 0.0. |
1.9.6