How to CCITT decode a blob - BlobToImage isn't doing it.

[Minok]

I'm getting a CCITT type 6 (group 4) compressed blob out of a CGM file, that represents the contents of a single tile in that format.
So its 920 bytes worth of CCITT compressed bitonal (B&W) data.

I'm trying to create an imagemagick image out of it but that process just fails and returns null, with error:
NoDecodeDelegateForThisImageFormat `' @ error/blob.c/BlobToImage/358.

Code: Select all

		MagickCoreGenesis(".", MagickTrue);	// initialize the imagemagick environment

		Image* imgMagickImage;			// an imagemagick image handle
		ImageInfo* imgMagickInfo= AcquireImageInfo(); // the imagemagick image info object
		ExceptionInfo* imgMagickException= AcquireExceptionInfo();	// where exception information is stored from calls

		imgMagickImage = BlobToImage(imgMagickInfo, (void*) dataBlob, (size_t)tileSize, imgMagickException);
		//Image* img = BlobToImage(const ImageInfo *image_info,const void *blob, 	const size_t length,ExceptionInfo *exception);

		CatchException(imgMagickException);	// test for and process imagemagick exceptions
		if (!imgMagickImage) { std::cerr << "ERROR: no magicImage returned - could not parse the blob" << std::endl; }

		MagickCoreTerminus();	// close & clean up the imagemagick environment

I've saved the blob to file and looked at it with a hex editor, and compared it with the blob that is stored in the CGM file from which it comes.
It is the same - that is, the blob is the CCITT compressed bit sequence for the compressed data.

Am I missing something here in how to use BlobToImage? When I fed it a blob that was a JPG encoded color image, that returned with the identified JPEG image.

Do I need to decorate the CGM sourced blob with some other stuff before I feed it to BlobToImage?

[Minok]

I've managed to find a solution to my problem and a guess a the reason.

Why doesn't it work as described above? I think ImageMagick does not support pure CCITT compressed data - but then it cannot possibly do so, because needed knowledge is not encoded in the CCITT blob such as the dimensions of the image, and a host of other things needed to correctly use the CCITT blob.

So what I did was, knowing the missing information, create a TIFF blob by adding the correct TIFF header and dictionary information, and then calling ImageMagic to convert the blob to an Image.

This amounted to adding the 4 bytes of type 42 intel encoding TIFF, along with an 11 entry TIFF Image File Directory in front of the blob that is the CCITT blob.
The tiff directory specified

• the tiff compression as 4 (CCITT Group 4)
• image width, height
• photometric mode as 0 (pixel value 1 is black, pixel value 0 is white)
• offset to the CCITT blob ( 154 bytes from the front of the fixed size header I have of 11 entries + storage for x/y resolution)
• samples per pixel = 1
• rows of image per strip = image height (ie the whole image in one row, all rows in the one strip)
• strip bytes count = how many bytes the strip is - ie the size of the CCITT blob
• resolution unit = 2 (inches) - may not be needed but TIFF docs seemed to suggest it
• X and Y resolution - an offset to storage of a rational number defining 100dpi in my case

Writing that produced tiff blob to the file system let me verify it was a valid tiff (as Windows would display that file correctly).

BlobToImage then returned and populated some meaningful information about the Image it created.