Added callback signal for filename of grabbed pcd file.

[hdino]

While grabbing pcd files it might be helpful to know which file is passed to the callback function. Use cases are (for example):

1. A device which provides point clouds in lower resolution than 2D images. Currently these 2D images have to be downsampled and stored within the point cloud. If you know which .pcd file is grabbed, one can store the 2D images next to the pcd data, e.g. in png files, and load them accordingly.
2. Sometimes it is necessary to store additional data, e.g. the exposure time, in addition to the point cloud itself. For that purpose one can create a csv file that corresponds the additional data to the filename of the point cloud.

[taketwo]

In my opinion, for the use-cases you listed it may make sense to implement a custom grabber. Of course, it would derive from the PCDGrabber and reuse most of the code. On top of that, it can deal with loading additional PNG, CVS, or whatever.

[hdino]

Yes, you're right, for the listed extraordinary use-cases it makes sense to implement a custom grabber. However, since the grabber's implementation is hidden in pcd_grabber.cpp, a custom grabber not compiled at the same time as the PCL wouldn't be able to access its internal parameters (like the currently grabbed file name). Thus, I think it would make sense to add a file-name-getter-method to the PCDGrabberBase's interface in order to avoid duplicated code or the necessity to build the PCL from source. Do you think this could be a practicable way?

[taketwo]

Makes sense. And since your changes do not break public API and do not introduce any noticeable overhead for those who are not interested in filenames, I'm in favor of merging this.

@jspricke what's your opinion?

[taketwo]

@jspricke I assume you just missed the notification. Can we get your opinion here?

[jspricke]

sorry, was just too far up in my list ;).