With reiterated thanks to Henrik for his code, I find that one must be careful with ImagePartition as the application of a function to the parts may result in an assembled image that is different than applying the function to the whole (unpartitioned) image.

Here's the unpartitioned code:

pageImage = ColorConvert[pageImage, "Grayscale"];
pageImage = ImageAdjust[pageImage, {2}];
pageImage = ColorToneMapping[pageImage];
pageImage = ImageAdjust[pageImage, {2, -.2, 1.5}];
pageImage = ColorNegate[Binarize[pageImage]];
Print[pageImage];

And here's the partitioned code (via Henrik):

pageImage = ColorConvert[pageImage, "Grayscale"];
partImgs0 = ImagePartition[pageImage, {200}];
partImgs1 = ParallelMap[ColorToneMapping@*ImageAdjust, partImgs0, {2}];
partImgs3 =  ParallelMap[ImageAdjust[#, {2, -.2, 1.5}] &,   partImgs1, {2}]; 
pageImage = ImageAssemble[partImgs3];
pageImage = ColorNegate[Binarize[pageImage]];
Print[pageImage];

See attached for original image and the partitioned and unpartitioned results.

All just FYI.

Cheers, Scott

Attachments:

Henrik ...

Thanks much for your comments and your code. Apologies for the delay in extending my thanks; I had to upgrade from 8 to 10 to understand the code. (Wolfram should credit you with the sale!) Cleaning followed by word extraction is indeed the objective. There is a tsunami of handwritten material being digitized. The time and expense of existing transcription methodologies coupled with declining population skills in cursive mean (IMHO, of course) we have to find new ways searching this material. In passing, I'm curious about the @* notation in line 4, in particular the *. Is this just function composition of some sort?

Thanks again.

Cheers, Scott

Thanks much, Matthew. Greatly appreciated. And thanks to you too, Eric, for your patience and attention to the matter.

Cheers, Scott

The image is the correct one. The numbers I noted are the number of black pixels in what comes out of ImageData:

pageImage = Import[inputFile];
pageImage=DeleteBorderComponents[pageImage];
binPage = MorphologicalBinarize[pageImage];
Pixels = ImageData[binPage];

In other words, DeleteBorderComponents turns the entire image black (1844 x 3069 = 5659236). Without DeleteBorderComponents there are 750132 black pixels in the image.

Cheers, Scott

Hello, Eric ...

Here's the code ...

pageImage = Import[inputFile];
pageImage=DeleteBorderComponents[pageImage];

I've attached the image.

Before DeleteBorderComponents I see 750,132 pixels. After DeleteBorderComponents I see 5,659,236.

Thanks in advance for any insight.

Cheers, Scott

Attachments: