|
FireSync API
|
Represents parameters for a camera tracking extension.
This extension enables the camera to automatically track the position of a laser line, dynamically adjusting the y-origin of the camera window in response to changes in the laser position. For acquisition systems with targets that remain relatively flat within the scope of a single camera frame, this can enable the use of a smaller ROI (yielding higher frame rates), without sacrificing overall measurement range across successive camera frames.
The tracking algorithm works by dynamically switching between search and tracking camera states. By convention, camera state 0 is considered the search state and camera state 1 is the first tracking state. These two states are required at a minimum; if additional states are defined, they are treated as extra tracking states.
In a typical tracking setup, the search state (camera state 0) is configured with a relatively large camera height value, so that the entirety of the measurement range can be observed in a single frame. The tracking state(s), in turn, are configured with smaller height values, enabling higher frame rates. Under ideal operation, the laser signal will be detected in a search frame and the tracking algorithm will then switch to tracking mode, dynamically adjusting the camera ROI y-position to keep the laser signal in the frame. The algorithm will remain in tracking mode indefinitely, only dropping back to search mode when the laser signal is lost.
The tracking algorithm proceeds from the search state to the first tracking state when the number of spots detected in the image is greater than or equal to the configured search threshold. The tracking algorithm will revert from tracking to search mode when the number of detected spots is less than the search threshold. If multiple tracking states are defined, the algorithm will always execute the complete set of tracking frames before switching modes or adjusting the ROI y-position. The determination about whether to switch modes is made after the final tracking frame is processed.
In current implementations, the window tracking algorithm is implemented through a combination of PL and CPU driver support. Accordingly, this extension requires a PL build with a column-based CG algorithm (e.g., Sobelv_Accgdw) and for the PL algorithm to be enabled. The PL algorithm performs bulk image analysis, while the CPU driver makes high level decisions about state changes.
Due to the time required for PL image processing and CPU decision-making, and the requirement to configure subsequent frames in advance of imager exposure, there is a 3-frame delay between when images are acquired and when corresponding tracking state changes are enacted.
Due to the use of CPU drivers, there is a non-negligible CPU run-time cost when this feature is used at high frame rates. At the time of this writing, current M1, M2, and M3 implementations have been established to work reliably at up to 5kHz. Higher frame rates may be possible (e.g., some tests have been successful at 7-8 kHz), but are not recommended in general. If run too fast, the consequence is likely to be dropped interrupts, with corrupted and/or out-of-sequence camera images as a secondary consequence.
Public Member Functions | |
| k32u | kCameraTracking_SearchThreshold (kCameraTracking extension) |
| Gets the tracking search threshold. More... | |
| kStatus | kCameraTracking_SetSearchThreshold (kCameraTracking extension, k32u threshold) |
| Sets the tracking search threshold. More... | |
 Public Member Functions inherited from kCameraExt | |
| kStatus | kCameraExt_Enable (kCameraExt extension, kBool enable) |
| Enables or disables the extension. More... | |
| kBool | kCameraExt_IsEnabled (kCameraExt extension) |
| Reports whether the extension is currently enabled. More... | |
Related | |
| #define | kCAMERA_TRACKING_STATE_SEARCH |
| Index of search state. | |
| #define | kCAMERA_TRACKING_STATE_TRACKING_0 |
| Index of first tracking state. | |
