problem with Grid/Collection Stitching
Locked
problem with Grid/Collection Stitching
 
|
Re: problem with Grid/Collection Stitching
|
|
Re: problem with Grid/Collection Stitching
|
|
Re: problem with Grid/Collection Stitching
|
Re: problem with Grid/Collection Stitching
|
|
Re: problem with Grid/Collection Stitching
|
|
Re: problem with Grid/Collection Stitching
|
|
Re: problem with Grid/Collection Stitching
|
|
Re: problem with Grid/Collection Stitching
|
|
Re: problem with Grid/Collection Stitching
|
|
I have been looking into the issue of the top left tile, because it
turns out that it is not just that the tile is not displayed. I attached a file (which I hope the listserv will pass) which shows the top row registered coordinates of a 13x18 mosaic. I stitched the following sizes: 13x6, 13x7, 13x8, 13x9, 13x10,13x11, 13x13, 13x15 and 13x18. The raw images are 1600x1200 pixels, and were acquired with 10% overlap. The strange thing is that the first tile is always at 0,0 in the list of registered locations. This would see reasonable, but note the y-coordinates of the subsequent images as the grid size increases. The y-coordinate of x=2, y=1, changes from -1.432 to -318.3 as the grid size changes from 13x6 to 13x18 (even though the top row should be unaffected. At first it is not noticeable, but beyond 13x10 it grows quickly. Note that the difference in the y-coordinate between x=13 and x=2 stays more or less constant (delta is around 18 or 19), but that first tile is pinned at (0,0), so it it moves out of the proper range for alignment. I can manually go into the coordinates file and put it in approximately the correct position, but it seems to me that there is some bug here. I hope this may help identify it. Now I understand why Yuekan needed that black row. Presumably the second row was ok and the black row did not matter. --aryeh On 09/10/2016 12:13, Aryeh Weiss wrote: > Following up on this thread -- I can add that the problem of the upper > left tile not appearing in the fusion occurs starting at 90 tiles. For > example, if I have a 10x12 layout, then the upper left tile will not be > displayed. If I do 10x9 of these tiles, the upper left tile will not be > displayed. But if I do 8x11, then it will be displayed. This behavior is > reproducible. > > The workaround described below works, even with defining a dummy row of > black images. > > --aryeh > > On 23/06/2016 5:32 AM, Aryeh Weiss wrote: >> Thank you for this reply. >> >> On 13/06/2016 9:50 PM, Yuekan Jiao wrote: >>> My method requires Grid/Collection type "Filename defined position", >>> i.e. >>> you will need to rename your images to the format of >>> "image_Y{yyy}_X{xxx}.tif". Then add a black row Y000 if your first >>> row is >>> Y001. Stitch all the images including Y000. >>> >>> I explored plugin Grid/Collection stitching a bit for the re-assembling. >>> Open the registration file "TileConfiguration.registered.txt", delete >>> the >>> lines of the added black row Y000. Run plugin Grid/Collection stitching >>> again, this time select type "Positions from file". Input >>> "TileConfiguration.registered.txt" >>> as "Layout file", uncheck "Compute overlap", you will get all the >>> images in >>> the stitched image. >>> Yuekan >> >> I finally had a chance to try this, and it works. I found that when I >> assembled the image this way, it was complete. >> This means that the problem was only in the display of the fusion, but >> the TileConfiguration.registered.txt file included the missing tile >> correctly, without the need for a black row. >> >> This is exactly what I needed to know. >> >> Best regards >> --aryeh >> >> >> >>> >>> >>> On Fri, Jun 10, 2016 at 7:41 AM, Aryeh Weiss <[hidden email]> >>> wrote: >>> >>>> Thank you to Michael and Yuekan for their replies to my inquiry. >>>> >>>> I have been traveling, but when I have a chance I will try to translate >>>> Michael's MATLAB script to an ImageJ script. >>>> >>>> I can add that I just tried to reassemble a 12 image grid which was >>>> named >>>> in similar way (tile_001, tile_002, ... tile_012), and >>>> which happened to be listed by row (the 120 image grid that showed this >>>> problem was ordered by column). >>>> It worked without leaving anything out. >>>> >>>> On 07/06/2016 8:57 PM, Yuekan Jiao wrote: >>>> >>>>> Hi, >>>>> >>>>> This problem was discussed a while ago, it looks that plugin >>>>> Grid/Collection Stitching still has it. What I do is to add a black >>>>> row >>>>> before the first row. After stitching all the images including the >>>>> added >>>>> black row, re-assemble the images without the added black row based >>>>> on the >>>>> registration file TileConfiguration.registered.txt. Either a script or >>>>> plugin Grid/Collection Stitching itself (just my impression) will do >>>>> the >>>>> job. I use my own script but have not tried plugin Grid/Collection >>>>> Stitching for the re-assembling. >>>>> >>>> Can you post your script for reassembling the montage? >>>> >>>> Yuekan >>>>> On Tue, Jun 7, 2016 at 8:23 AM, Majurski, Michael Paul (Fed) < >>>>> [hidden email]> wrote: >>>>> >>>>> Is the upper left image (0,0) missing? Or is it misplaced? >>>>>> I have had this happen, but looking in the >>>>>> TileConfiguration.registered.txt file shows the (0,0) image has >>>>>> simply >>>>>> been >>>>>> put in the wrong place, ending up under another image using an >>>>>> overlay >>>>>> blend. >>>>>> >>>>> My TileConfiguration.registered.txt shows the first image as being at >>>> (0.0, 0.0), which I assume makes sense because everything else moves >>>> relative to that image. >>>> >>>> For assembling the TileConfiguration.registered.txt file I use a Matlab >>>>>> script to build the stitched image. >>>>>> If you would like a copy of my script send me an email (its long >>>>>> enough I >>>>>> did not want to include the whole thing in the email body). >>>>>> >>>> As noted above, I will try to adapt that script to ImageJ and I will >>>> try >>>> it. >>>> >>>> >>>> ~Michael Majurski >>>>>> [hidden email] >>>>>> >>>>> Best regards >>>> --aryeh >>>> >>>> >>>> -----Original Message----- >>>>>> From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of >>>>>> Aryeh Weiss >>>>>> Sent: Tuesday, June 07, 2016 10:07 AM >>>>>> To: [hidden email] >>>>>> Subject: problem with Grid/Collection Stitching >>>>>> >>>>>> I am not sure if this was posted, so I am resending. My apologies >>>>>> if was >>>>>> already posted. >>>>>> >>>>>> >>>>>> I have a directory with files named >>>>>> bare60C12minFlat000.tif. bare60C12minFlat001.tif, >>>>>> bare60C12minFlat002.tif, .... >>>>>> >>>>>> So in the plugin I define; >>>>>> First file index = 0 >>>>>> File names for tiles = bare60C12min{iii}.tif >>>>>> >>>>>> And the plugin does a great job for 120 images, except that the upper >>>>>> left corner (0,0) in the fused image is left out, and is black. >>>>>> >>>>>> The first image in teh ROI list is bare60C12minFlat001.tif . For >>>>>> example, here is the first ROI name: >>>>>> sp=1; label=0; series=0; C=1; Z=1; T=1; file=bare60C12min001.tif-0 >>>>>> >>>>>> I tried renaming the entire directory to start at 001, but then I get >>>>>> the first file as: >>>>>> sp=1; label=0; series=0; C=1; Z=1; T=1; file=bare60C12min002.tif-0 >>>>>> >>>>>> and so on. >>>>>> >>>>>> In there a way to get the plugin to include that first file? >>>>>> >>>>>> Also, given the TileConfiguration.registered.txt file, is there a >>>>>> way to >>>>>> reconstruct the fused image, or should I write a script that reads >>>>>> the >>>>>> coordinates, inputs the images, and pastes them appropriately? >>>>>> >>>>>> Thanks in advance. >>>>>> >>>>>> --aryeh >>>>>> >> > > -- Aryeh Weiss Faculty of Engineering Bar Ilan University Ramat Gan 52900 Israel Ph: 972-3-5317638 FAX: 972-3-7384051 -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
topRowStitching.txt (10K) | Free forum by Nabble | Edit this page |