How to convert the raw hyperspectral image (hawaii or yellowstone)into .mat file?

I have raw images which I has taken from the link

https://aviris.jpl.nasa.gov/data/free_data.html

https://coding.jpl.nasa.gov/hyperspectral/

I need to convert this raw files to .mat file through the matlab, so that I can get the image band by band

Answer 1

[Walter Roberson]

X = multibandread('hawaii_sc01.raw', [512, 614, 224], '*uint16', 0, 'bip', 'ieee-be');

The datatype is uint16 but the data is 12 bit, so the used portion is theoretically 0 to 4095 but in practice the range is 82 to 1006, so call it [0 1023]

According to the README, that particular file is the above size, and the maine file below it is a different size also 12 bit, and all of the rest of the files are 512 lines x 680 samples x 224 bands, instrument bit depth = 16 bits which would correspond to

X = multibandread(filename, [512, 680, 224], '*uint16', 0, 'bip', 'ieee-be');

You would have to examine the values to see what the actual data range is.

[Shrish Bajpai]

If I i use the command

X = multibandread(filename, [512, 680, 224], '*uint16', 0, 'bip', 'ieee-be');
X1=X(:,:,10);
imshow (X1,[])

I am unable to get the image what i should get

[Walter Roberson]

The data is there properly, at least for the uncalibrated aviris_sc0.raw

imshow() uses grayscale as the default colormap. The color samples they show are pseudocolor applied to calibrated (not raw) images; I do not know which colormap they are using or how they are blending the bands.

The used data range for X1 is 3386 to 13512. However, not much of it is on the higher end. If you do the imshow() like you do, and then use imcontrast and use the widget to automatically do outlier rejection at 2%, then then automatic upper bound is on the order of 6000

Some of the other slices actively use higher values. Slice 128 actively uses up to about 8000. Slice 46 actively uses up to about 12386 (exact position arguable) with peak over 21000.

[Shrish Bajpai]

Respected Sir,

As per ypur suggestion, I used the following commands to draw the i image of the band 100, but I am not getting what I need to get get. It is image that contains only alt and pepper noise.

X = multibandread('aviris_sc0.raw', [224 680 512],'int16',0, 'bip', 'ieee-be');
X100=squeeze(X(100,:,:));

I request you kindly guide me if I am wrong.

[Walter Roberson]

filename = 'aviris_sc0.raw';
X = multibandread(filename, [512, 680, 224], '*uint16', 0, 'bip', 'ieee-be');
X100 = fliplr(X(:,:,100));
imshow(X100,[])

Now compare to the first false-color image at https://coding.jpl.nasa.gov/hyperspectral/falsecolor.html and it will be obvious that the data has been extracted in the right order.

Answer 2

[Shrish Bajpai]

Respected Sir,

It is working!!!

Thanks for guiding me and solving my problem which I was stuck for last six month (still hawaii is there which I am not able to convert).

Again I thanks to you by my folding hands (Indian Hindu culture) for guiding me on the issue.

Regards,

Shrish Bajpai

Detail Code as follows

Dim = [512,680,224]; % lines, samples, bands

dType = 'int16'; % for data type 2

hOffset = 0;

intlve = 'bip';

border = 'ieee-be'; % For byte order 1

X = multibandread('aviris_sc3.raw',Dim,dType,hOffset,intlve,border);

figure; imshow(X(:,:,1),[])

[Walter Roberson]

Earlier I posted

X = multibandread('hawaii_sc01.raw', [512, 614, 224], '*uint16', 0, 'bip', 'ieee-be');

Before I posted that I tested it and compared it to the sample images that they provided, so I was sure that the data was read properly.