1

Topic: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Sir
Could you please help me?
I need to validate my work (Monitoring migration rates of sand dunes) using COSI-CORR.
I got 2 SPOT images for the same area, (2007 SV basic ortho 2.5m 3 band) and (1995 2 P Level 2A 10m) resized to 2.5m
I have finished the mosaicking and geometric corrections using ERDAS and Arc GIS, now I have the final images with only *.tif extension , with the same pixel size and subset (Columns 360, Rows 461) for small area with just 2 individual sand dune. I tried too much to start directly from correlation presses but nothing happed error with images or wrong output
also I tried to start from the first process (Ancillary Spot) with my row-data to follow the extensions but the program doesn’t accept my data (1995 METADATA.DIM) and give me massage after processing (invalid widget identifier: 8160) and for spot 2007 (point type required in this context:TREE)
So please help me how I can use your program with my images, I need a simple case study
Please Sir I am waiting your answer as soon as you can

My best regards
Ghadiry

2

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Ghadiry,

Several things:

1- If you want to insure accurate processing and co-registration, you should use L1A images. Most likely the mosaicing and geometrical processing you applied won't be of sufficient quality.

2- Starting from "the first process" requires the raw data with raw metadata (only available in L1A format). This is why you cannot do the whole processing, because you don't have the raw data (hence the error when generating the ancillary file).

3- An overlap between your 2 images of only 360 by 461 pixels is extremely small and will not allow you to estimate the geometrical errors of your processing. Images are large, use the largest possible overlapping area.

4- Images having the same pixel size is insufficient for correlation. The ortho-grid defined should also exactly overlap. Let us know what the error message is in the correlation and what parameters you are using.

5- It is wrong to compare the images at a resolution of 2.5m. You should compare the 2 images at the lowest common resolution, i.e. 10m.

Hope this help, let us know.

Best,
Sebastien

3 (edited by mghadiry 2010-03-10 13:53:51)

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Sir

I would like to thank you very much for your prompt answer

I have changed everything, 2 images with 10m resolution panchromatic (Columns 1397, Rows 1311) for both image with mask for the same area exactly

I try with Correlation command but I got this massage (images cannot be overlapped due to their origin and resolution)

I have used default parameters in frequency options. Is it true? If it’s wrong could you please tell me how I can adjust frequency options with my images?

Sincerely
Ghadiry

4

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Gahdiry,

You get this error message because the two ortho-images are not aligned, even though they have the same resolution. Hence the correlation will measure the grid misalignment of the grids, which we don't allow. You can either regenerate your ortho-images to make sure they are defined on exactly the same grids, or you can use a trick in ENVI: Basic Tools -> Layer stacking, output file exclusive and select cubic convolution resampling. Then you'll have a file with the two images that belong to the same exact grid.

You should read the COSI-Corr manual to make sure the default parameter for the frequency correlation are well chosen for the process you're trying to measure. You can also have an idea of how well your images are registered by opening 2 displays, then link the displays and look at the differences.

Sebastien

5

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Sebastien

Thanks a lot for your help

1- As you told me I have used layer stacking for the both Spot images 1995-2007(1397 x 1311). I tried again with Correlation and I got error message (images coordinates origins must be a multiple of the resolution for a grided output). How I can dissolve this problem?

2- I switched off (Grided Output) from Frequency Options and its succeed but I got very small output (84 x 78).
- How I can get the same input size?
- How I can switch on (Grided Output) without the last error message?

Sincerely
Ghadiry

6

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

You have to use a special orhho-grid which is generated only by COSI-Corr to use the grided correlation. It generates  grid which will be common to many independent correlations for better comparaison if you study time series. In your case, you have to uncheck the button.

the output size is determined by the sampling of the correlation, determined by the correlation step size. Please refer to the COSI-Corr manual.

Cheers,
Sebastien

7

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Sebastien

1-  As you know I have problem with my images so I can’t start from Ancillary analysis to have orhho- grid, so I switched off from correlation I got good results but the output pixelizd (322 x 300) and the original file (3265 x 3045). I tried many times to dissolve this problem but nothing happen please if you have solution for this problem inform me.

2- I have out nice output (RGB) color, so how I can plot the image legend including the color classes which is the movement rates of sand dunes

Sincerely
Ghadiry

8

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Ghadiry,

1- I understand you cannot do the complete process as the images you're working with are already ortho-rectified. However I do not understand your question #1, could you please rephrase it?

2- The correlation produces a 3-band file. The first two bands represent the projection of the horizontal displacement vector along EW and NS direction. The last band is the SNR. You shouldn't open the bands in RGB, but each band individually and study each band individually. If you want an estimate of the displacement, you can compute the displacement norm, but be careful of the noise impact, see this paper:
http://www.tectonics.caltech.edu/slip_h … u_2009.pdf

Best,
Sebastien

9

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Sebastien

Really I would like to thank you for your prompt replay,

1- My correlation results it’s too small, size (322 x 300) it’s not like input file (3265 x 3045). Do you have solution form this problem or this small image size is normal by COSI-CORR?

2-I already extract the vector layer with movement rates and direction from the correlation process, and I plot the results in the layout with grid and scale bar, but how I can drag the legend of image reflectance ?(the colors RGB its already reflected the rates of the dunes movement).

Sincerely
Ghadiry

10

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Dear Ghadiry,

1- As explained in one of my previous message, the sampling of the correlation is determined by the correlation step size. This is well explained in the user's manual. Modify the step size to change the output size.

2- "drag the legend of image reflectance"? I'm sure i understand this. If you want to select different colormaps for your data set, in the current display, go to: Tools -> Color Mapping -> Envi Color Tables...

Sebastien

11 (edited by mghadiry 2010-03-18 13:55:28)

Re: Spot 5 basic ortho & Spot 3 2A image Ancillary creation

Thanks you Sir

Sorry for inconvenience

Ghadiry