ocr.js

const Tesseract = require('tesseract.js');
const sharp = require('sharp');
const cv = require('opencv4nodejs');
const fs = require('fs');

const EROSION_SIZE = 1;
const EROSION_MODE = cv.MORPH_ERODE;
const SCALE = 2;
const ITERATIONS = 2;
const ANCHOR = new cv.Point2(EROSION_SIZE, EROSION_SIZE);

const dirs = fs.readdirSync('captchas');

function runTesseract() {
  return Tesseract.recognize('output_edited_opencv.png', {
    tessedit_char_whitelist: '0123456789',
    tessedit_char_blacklist: ' ',
    tessedit_pageseg_mode: 5,
    tessedit_ocr_engine_mode: 1,
  }).then(result => console.log(String(result.text).replace(/[\s\n\r]/ig, '')));
}

sharp(`./captchas/${dirs[6]}`)
  .negate()
  .removeAlpha()
  .threshold(50, { grayscale: true })
  .toFile('output_edited.png')
  .then(() => cv.imreadAsync('output_edited.png'))
  .then(img => img.resizeAsync(
    SCALE * img.sizes[0],
    SCALE * img.sizes[1],
    2, 2,
    cv.INTER_NEAREST,
  ))
  // .then(img => img.erodeAsync(new cv.Mat(), ANCHOR, ITERATIONS, EROSION_MODE))
  .then(img => img.morphologyExAsync(new cv.Mat(), cv.MORPH_OPEN, ANCHOR, ITERATIONS, EROSION_MODE))
  .then(img => cv.imwriteAsync('output_edited_opencv.png', img))
  .then(runTesseract);