update ply_io and filter_branch.py

[wanxinyang]

ply_io.py

1. .ply exported from CloudCompare has "scalar_" in property names, so add a if statement to remove "scalar_" and get expected names.
2. If there is NAN values in any of the properties, read_ply() fails. So add a step to remove NAN values when read data from ply files.

filter_branch.py

1. Currently, qrdar library reports some error when import it, so I change to import read_ply() and write_ply() from ply_io.py instead of qrdar.io
2. Add reflectance filtering [-20, 5] in case of RiEGL VZ400i records values down < -30dB.

[philwilkes]

Instead of adding this section I would change line 28 to prop.append(line.split()[2].replace('scalar_', ''))

[philwilkes]

Why are there NaN values?

Also this below is slightly more succinct?

    if fmt == 'binary':
        arr = np.fromfile(ply, dtype=','.join(dtype))
        df = pd.DataFrame(data=arr, columns=prop)
    else:
        df = pd.read_csv(ply, sep='\s+', names=prop, header=None)
       
    df = df.dropna()  # drop rows with NAN values

[wanxinyang]

The last point of an extracted cloud has NAN values in the last two columns (it does not have the last two properties), which caused filter_branch.py stopped running.

Yes, I agree. Moving df.dropna() out of the if statement is more succinct. Thanks.

[philwilkes]

OK, how is the cloud being written in the first place before being read here as I think there must be an underlying issue. Dropping NaN here is a fix but could hide other errors e.g. what if it was the last 10,000 lines? I think I would put this outside the ply_io function.

[Philip035]

The qrdar library still can't be imported now.