Third-party sensors are input devices, such as 3D sensors and scanners, that acquire image components, such as depth maps (range images), intensity images, and confidence maps. To support various manufacturer's 3D devices, Matrox Design Assistant uses a third-party 3D sensor platform component (specified in the Cameras page of the Platform Configuration dialog). For supported devices, the platform component implements retrieval of the various output image components (at minimum, a single range file) via the 3D sensor's SDK or API, and provides them to the Matrox Design Assistant Camera step. The exact image components available and their data types are sensor-dependent. Note that the 3D sensor platform component only handles data retrieval. All other configurations of the device are done using the interactive software provided by the sensor manufacturer.
If your third-party sensor is not already supported in Matrox Design Assistant, contact your Matrox sales representative.
When working with a depth map acquired from a 3D sensor, the depth map is typically already calibrated in Z, giving depth values that correspond to real-world distances from the sensor's reference position. Matrox Design Assistant processing and analysis steps use real world Z-values for inputs such as threshold levels when working on depth maps.
The 3D sensor might also provide world X- and Y-coordinates. If not, you can apply a calibration, but only a scale type of calibration is available for depth maps. To apply a calibration, specify the calibration in the Camera step. For more information, see the Calibration section in Chapter 30: Calibration.
For depth maps, invalid values are an important concept. Often, not all pixels in a depth map will have a true depth value, due to occlusion and shadowing of the sensing beam. Spurious reflections (for example, from rough, shiny surfaces) can also result in invalid values. Depending on the manufacturer, invalid pixels are identified by a special value (for example, a maximum value or zero), or by a separate mask or map (confidence map). A neighboring group of invalid pixels is sometimes called a gap. For more information on gaps and how to handle invalid values, see the Filling gaps subsection of the Procedure for using a depth map section in Chapter 38: Working with depth maps.