Improve QR Code Recognition from Image

findBarcode.py

from flask import Flask, jsonify, request
from pyzbar.pyzbar import decode
from pyzbar.pyzbar import ZBarSymbol

import cv2
import numpy as np
import base64
import urllib
import pdf2image

app = Flask(__name__)

@app.route("/images/qrcode", methods=["POST"])
def detect_qrcode():
  body = request.get_json()
  app.logger.info(f"Received QR code detection request")

  # parse and decode input from data URI
  with urllib.request.urlopen(body["payload"]) as response:
    if response.headers['Content-Type'] == "application/pdf":
      # NOTE Increase `dpi` can improve detection, but slower
      pages = pdf2image.convert_from_bytes(response.read(), dpi=500, single_file=True)
      image = cv2.cvtColor(np.array(pages[0]), cv2.COLOR_RGB2BGR)
    else:
      image = np.asarray(bytearray(response.read()), dtype=np.int8)
      image = cv2.imdecode(image, cv2.IMREAD_COLOR)

  # attempt with zbar first, preprocess may not be required
  result = decode_qrcode(image)
  qrcode = None

  # crop the qr code region and decode with zbar
  if not result or not len(result):
    qrcode = find_qrcode(image)
    result = decode_qrcode(qrcode)

  app.logger.info(f"Decoded QR code: {result}")

  return jsonify({
    # decoded qr code payload, supports extracting multiple qr codes
    "data": [d.data.decode("utf-8") for d in result],
    # original image with highlighted qr code region
    "image": base64_image_uri(image),
    # preprocessed qr code cropped from the original image
    "qrcode": base64_image_uri(qrcode),
  })

def find_qrcode(image: np.ndarray):
  source = image.copy()

  # Load imgae, grayscale, Gaussian blur, Otsu's threshold
  gray   = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
  blur   = cv2.GaussianBlur(gray, (9,9), 0)
  thresh = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

  # Morph close
  kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5,5))
  close  = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel, iterations=2)

  # Find contours and filter for QR code
  cnts = cv2.findContours(close, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
  cnts = cnts[0] if len(cnts) == 2 else cnts[1]
  for c in cnts:
    peri    = cv2.arcLength(c, True)
    approx  = cv2.approxPolyDP(c, 0.04 * peri, True)
    x,y,w,h = cv2.boundingRect(approx)
    area    = cv2.contourArea(c)
    ar      = w / float(h)
    m       = 2 # NOTE add margin to the cropped image
    if len(approx) == 4 and area > 1000 and (ar > .85 and ar < 1.3):
      cv2.rectangle(image, (x, y), (x + w, y + h), (36,255,12), 3)
      qrcode = source[y-m:y+h+m, x-m:x+w+m]
      break

  # Enlarge the cropped qr code to improve detection
  return cv2.resize(qrcode, [500, 500], interpolation = cv2.INTER_CUBIC)

def decode_qrcode(image: np.ndarray):
  return decode(image, symbols=[ZBarSymbol.QRCODE])

def base64_image_uri(image: np.ndarray):
  if image is not None and len(image) > 0:
    encoded = base64.b64encode(cv2.imencode('.jpg', image)[1]).decode("utf-8")
    return f"data:image/jpg;base64,{encoded}"

Quick Start

Install the required dependencies:

pip3 install flask kraken pyzbar Pillow opencv-python pdf2image

Launch the demo application:

export FLASK_APP=main.py
export FLASK_ENV=development

flask run --host=0.0.0.0

Extract QR code from an image file:

curl -X POST 'http://localhost:5000/images/qrcode' \
     --header 'Content-Type: application/json' \
     --data-binary '{ "payload": "file:///tmp/qrcode.png" }' |
    jq -r '.data[0]'

Extract QR code from a base64 encoded image:

base64 /tmp/qrcode.png | xargs printf '{ "payload": "data:image/png;base64,%s" }' |
    curl -X POST 'http://localhost:5000/images/qrcode' \
         --header 'Content-Type: application/json' \
         --data-binary @- |
    jq -r '.data[0]'