Tesseract.js version (version number for npm/GitHub release, or specific commit for repo)
5.1.0

Describe the bug
I have processed an image to monochromatic black and white with no noise. I then tell tesseract.js to OCR the top half. Tesseract.js OCRs the bottom half, and interprets blocks of pure white as tildes

To Reproduce
Steps to reproduce the behavior:
(async () => {
const worker = await createWorker('eng', 1, {
/* logger: (m) => nodecg.log.info(m) */
});
worker
.recognize(__dirname + '/OBS.png', { rectangle: { top: 0, left: 0, width: 350, height: 39 } })
.then((result) => {
fs.writeFileSync(__dirname + '/ocr.json', JSON.stringify(result));
console.log(result.data.text);
console.log('Confidence: ' + result.data.confidence);
});
})();

Please attach any input image required to replicate this behavior.

Expected behavior
I expect this run on the attached image to result in "WWWWWWWWWW", or at least "Mark", not "~~ Mark"
Hopefully I'm doing something wrong and this is not actually a bug?

Device Version:

• Windows 10
• Node 18.19.0

Additional context
Add any other

context about the problem here.