I was attempting to make catalog for one of the Frontier-fields cluster. But I realized some faint objects are not in my catalog.

I saw Segmentation map and realized this object did not consider as true clump, while, the fainter objects considered as a clump.

When I run Segment with --checksegmentation and --checksn, I expected number of difference between the clump’s peak value and the highest valued river pixel around that clump when divided by the Sky standard deviation under the river pixel, for main objects become smaller than the faint objects which considered as clumps.
Because the criteria to select true from false clumps is the peak significance.

Unfortunately, this value was approximately similar for main clump and fainter object's clump and both of them should not consider as true clump. I do not know where is problem.

I think this is a bug. For producing this result and check problem, I write my complete commands in below.

At first, I downloaded a frontier-fields cluster by below command:

wget https://archive.stsci.edu/pub/hlsp/frontier/abell2744/images/hst/v1.0/hlsp_frontier_hst_wfc3-60mas_abell2744_f160w_v1.0_drz.fits -Oabell2744-f160w.fits

Secondly, based on the below command I crop the deeper part:

astcrop abell2744-f160w.fits -h0 --polygon="3.6034055,-30.4165889:3.5655321,- 30.4135259:3.5722988,-30.3765434:3.6103451,-30.3816284" --mode=wcs --output=crop.fits

Then I run NiseChisel on output of the astcrop:

astnoisechisel crop.fits --output=nc.fits

Finally, I have make a flat(3*3) kernel and based on this kernel Segment was built.

astconvert flat.txt --output=kernel.fits
astsegment nc.fits --kernel=kernel.fits --snquant=0.99 --gthresh=1000 --grownclumps --output=seg.fits

I would be thankful if you check this problem.