MrSid file display problem in 10.1

[GregBacon]

Hello.
I am having a problem with the display of a Generation 3 MrSid file in ArcGIS 10.1 that I didn't have in 9.3.1 or 10.0.  The circled areas in the attached image should not display as NoData as there is imagery here. 

According to the 10.1 Help, Generation 4 MrSid files and all previous versions are supported so I don't know what else this could be. 

Any help is appreciated.

[ATTACH=CONFIG]20749[/ATTACH]

[larryzhang1]

Hello.
... The circled areas in the attached image should not display as NoData as there is imagery here.  ...

[ATTACH=CONFIG]20749[/ATTACH]

Firstly, identify the DN (pixel values) inside the circled areas by �??Identity�?? tool and see if those are �??0�?? value.

If yes, that indicates that both the background and circled areas in the MrSID imagery share same value (�??0�??) by default in ArcGIS, even though those are different from imagery perspective.

So, use one of the following options, so that ArcGIS can properly differentiate those circled areas of the MrSID imagery against the background:

1. Under �??Symbology�?? of the Layer Properties in ArcGIS, check �??Display Background Value�?? and setup any color (saying, red); meanwhile, change default color in �??Display NoData�?? to other (saying, black). And then see if it works. If not, use the following solution to ensure that this issue will be solved permernently;

2. Import this MrSID into ERDAS (MrSID licensed) and then reprocess it via the Histogram/LUT. Make sure that the DN values inside those circled areas are over �??0�?? (saying, �??1�??) and then save as new image in IMG (or, export into MrSID again, if preferable)

[GregBacon]

The values in the circled areas are not, and never were, zero (unless you look at them in ArcMap 10.1).  ArcMap 9.3.1 and 10.0 as well as ERDAS 2010 recognize this; all areas of the image display properly in these applications (see attached).   

I'm trying to figure out why ArcMap 10.1 is the only application that incorrectly displays these areas with a value of zero.  I should add that I have installed Service Pack 1 and the problem remains.

  [ATTACH=CONFIG]20794[/ATTACH]

[EricRice]

Greg,

Zoom all the way in on one of those circled areas.  Does the image look normal?  I'm trying to determine if a pyramid level is bad or if the source pixel is actually black.  Your screenshots are zoomed out enough that we are surely looking at a pyramid and not the source.  This could be a defect that needs reported to Esri Support.  You don't need maintenance to submit a defect to Support just in case you don't have maintenance.

Best,
Eric

[GregBacon]

Eric,

The black areas still appear when I zoom in and the Identify tool still returns a pixel value of zero.  I suspected that it might be the pyramids and would have deleted them to start over if they weren't internal to the MrSid format.

Thanks for your help.

[larryzhang1]

... the Identify tool still returns a pixel value of zero. ...

From 'imagery processing' perspective, most likely, pixel values of zero inside the 'processed' raster imagery are not valid (except for DEM, and a few of cases for specific purposes).

In other word, the pixel values in �??processed�?? panchromatic imagery should be ranged from 1 to 254 for 8-bit products. (Of course, 1-255 are widely acceptable). For RGB imagery, similarly, the pixel values (0, 0, 0) also should be avoided in the deliverables of 'processed' imagery.

In fact, through this kind of inspection onto �??processed�?? imagery inside ArcGIS, it has been became one of the easiest practices as common QA/QC exercises in operation (such as, to avoid voids and black edge, etc.).

[GregBacon]

Eric,

Is there a link on the Support page to submit a defect? 

-Greg

[EricRice]

Greg,

Click the link that says "Support Request Form", "Chat with an Analyst" or just phone in. 

Best,
Eric