Corners detection from Point Cloud is wrong.

[DinoBN]

<< Description >>
after run the code of "6.Non-linear optimization for final extrinsic parameters."
output data 0001_cal_backproj.png 〜 0004_cal_backproj.png is obviously wrong.

"img_corners_est" is successfully corners detected.
but, the corners from point cloud data doesn't match the corners from img data.

Do I need to change config.yaml from the default value to work with the data "ILCC_sample_perspective_data"?

If so, please let me know the setting value.

<< Details >>
I installed　the most recent version ILCC
and run with the sample data "ILCC_sample_perspective_data".

about config.yaml
I changed the below parameters.
'base_dir' : the data directory
'poses_num' : 20 → 4 (there are only 4 data at the sample data)

After the command "LM_opt.cal_ext_paras()".
the contents of "YYYYMMDD_HHMMSS_calir_result.txt" is below.

4.715408122537658286e+00
2.337674434525445211e+00
1.106741906490246730e-01
9.065658045971332069e-01
2.508680940249127245e-01
-2.121439663123081965e-01

<< Verification >>

I tried only the code "After the aforementioned process, utility module can be imported for visualizing various of results. "
with "ILCC_sample_perspective_data"

and the output data 0001_cal_backproj.png 〜 0004_cal_backproj.png is well.
After the command "LM_opt.cal_ext_paras()".
the contents of "YYYYMMDD_HHMMSS_calir_result.txt" is below.

1.338479622733038943e-01
-1.524339677023719419e+00
1.441150134721521425e+00
-6.890869881762318530e-03
-1.935656845054284925e-01
-3.367990535562562227e-01

<< My Environment >>
OS : Ubuntu16.04 64bit
ILCC : most recent version ILCC (now 2019/1/10)
OpenCV : 3.3.1-dev
MATLAB : 9.5.0.944444(R2018b)
PCL : 1.7
VTK : 8.1.1
python : 2.7.15

[mfxox]

@DinoBN Thank you for your detailed report.
As you mentioned in the estimated pcd corner figure with the utility.vis_ested_pcd_corners() command, the corner seems not correctly estimated.

A well estimated corner result should be visualized like the following, blue and red points are arranged like chess pattern. And there is no chess-pattern like in your result.

From this point, we can assume that the column of the intensity is not set correctly.
In the sample data, the intensity column of the point cloud csv file is 3 (counting starts from 0), so you need to change intensity_col_in to 3 in config.yaml .