Grid Image CAPTCHA: Mapeo de coordenadas y selección de celdas

Los CAPTCHA de imágenes de cuadrícula, como los desafíos de imágenes reCAPTCHA v2, presentan una cuadrícula de 3 × 3 o 4 × 4 y piden a los usuarios que seleccionen celdas que coincidan con una instrucción ("Seleccione todos los cuadrados con semáforos"). CaptchaAI devuelve los índices de celda. Esta guía cubre cómo capturar la cuadrícula, asignar celdas a coordenadas y hacer clic en los mosaicos correctos.

Diseños de cuadrícula

Los CAPTCHA utilizan dos tamaños de cuadrícula estándar:

3×3 Grid:          4×4 Grid:
1  2  3            1   2   3   4
4  5  6            5   6   7   8
7  8  9            9  10  11  12
                   13  14  15  16

Las celdas están numeradas de izquierda a derecha y de arriba a abajo (orden de lectura).

Paso 1: captura la imagen de la cuadrícula

Pitón (selenio)

import base64
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC

driver = webdriver.Chrome()
driver.get("https://example.com/form")

# Wait for reCAPTCHA iframe
WebDriverWait(driver, 10).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "iframe[src*='recaptcha']"))
)

# Switch to challenge iframe
iframes = driver.find_elements(By.CSS_SELECTOR, "iframe[src*='recaptcha']")
challenge_iframe = iframes[-1]  # Challenge iframe is typically the last one
driver.switch_to.frame(challenge_iframe)

# Get the grid image
grid_img = driver.find_element(By.CSS_SELECTOR, "img.rc-image-tile-33, img.rc-image-tile-44")
img_src = grid_img.get_attribute("src")

# Get instruction text
instruction = driver.find_element(
    By.CSS_SELECTOR, ".rc-imageselect-desc-wrapper"
).text
print(f"Instruction: {instruction}")

# Screenshot the grid as base64
img_b64 = grid_img.screenshot_as_base64

# Determine grid size
classes = grid_img.get_attribute("class")
grid_size = "4x4" if "44" in classes else "3x3"
print(f"Grid size: {grid_size}")

driver.switch_to.default_content()

JavaScript (titiritero)

const puppeteer = require('puppeteer');
const fs = require('fs');

const browser = await puppeteer.launch({ headless: false });
const page = await browser.newPage();
await page.goto('https://example.com/form');

// Find the challenge iframe
const frames = page.frames();
const challengeFrame = frames.find(f => f.url().includes('recaptcha'));

// Get instruction
const instruction = await challengeFrame.$eval(
  '.rc-imageselect-desc-wrapper',
  el => el.textContent.trim()
);

// Screenshot the grid image
const gridImg = await challengeFrame.$('img.rc-image-tile-33, img.rc-image-tile-44');
const imgBuffer = await gridImg.screenshot();
const imgBase64 = imgBuffer.toString('base64');

// Determine grid size
const className = await challengeFrame.$eval(
  'img.rc-image-tile-33, img.rc-image-tile-44',
  el => el.className
);
const gridSize = className.includes('44') ? '4x4' : '3x3';
console.log(`Grid: ${gridSize}, Instruction: ${instruction}`);

Paso 2: enviar a CaptchaAI

import requests
import time
import json

API_KEY = "YOUR_API_KEY"

# Parse the instruction to a simple keyword
# "Select all images with traffic lights" → "traffic lights"
import re
keyword_match = re.search(r'(?:with|of|containing)\s+(.+?)\.?$', instruction, re.I)
keyword = keyword_match.group(1) if keyword_match else instruction

# Submit
with open("/tmp/grid.png", "wb") as f:
    f.write(base64.b64decode(img_b64))

with open("/tmp/grid.png", "rb") as f:
    resp = requests.post("https://ocr.captchaai.com/in.php", 
        files={"file": f},
        data={
            "key": API_KEY,
            "method": "post",
            "grid_size": grid_size,
            "img_type": "recaptcha",
            "instructions": keyword,
            "json": "1",
        }
    ).json()

if resp["status"] != 1:
    raise Exception(f"Submit error: {resp['request']}")

task_id = resp["request"]

# Poll
for _ in range(20):
    time.sleep(5)
    result = requests.get("https://ocr.captchaai.com/res.php", params={
        "key": API_KEY, "action": "get", "id": task_id, "json": "1"
    }).json()

    if result["status"] == 1:
        cells = json.loads(result["request"])
        print(f"Cells to click: {cells}")  # e.g., [1, 3, 6, 9]
        break
    if result["request"] != "CAPCHA_NOT_READY":
        raise Exception(f"Error: {result['request']}")

Paso 3: Asigne índices de celda para hacer clic en coordenadas

Convierta índices de celda basados en 1 a coordenadas de píxeles dentro de la cuadrícula:

def cell_to_coordinates(cell_index, grid_size, grid_width, grid_height):
    """Convert a 1-based cell index to (x, y) center coordinates."""
    if grid_size == "3x3":
        cols, rows = 3, 3
    else:
        cols, rows = 4, 4

    cell_w = grid_width / cols
    cell_h = grid_height / rows

    # Convert 1-based index to 0-based row/col
    idx = cell_index - 1
    col = idx % cols
    row = idx // cols

    # Center of the cell
    x = col * cell_w + cell_w / 2
    y = row * cell_h + cell_h / 2

    return int(x), int(y)

# Example: grid is 300×300
for cell in cells:
    x, y = cell_to_coordinates(cell, grid_size, 300, 300)
    print(f"Cell {cell} → ({x}, {y})")

Salida para una cuadrícula de 3×3 (300×300):

Cell 1 → (50, 50)
Cell 3 → (250, 50)
Cell 6 → (250, 150)
Cell 9 → (250, 250)

Paso 4: haz clic en las celdas

selenio

from selenium.webdriver.common.action_chains import ActionChains

driver.switch_to.frame(challenge_iframe)

# Get grid element position and size
grid_el = driver.find_element(By.CSS_SELECTOR, ".rc-imageselect-target")
grid_rect = grid_el.rect
grid_w = grid_rect["width"]
grid_h = grid_rect["height"]

actions = ActionChains(driver)

for cell in cells:
    x, y = cell_to_coordinates(cell, grid_size, grid_w, grid_h)
    # Click relative to grid element's top-left corner
    actions.move_to_element_with_offset(
        grid_el,
        x - grid_w / 2,  # offset from center
        y - grid_h / 2
    ).click()

actions.perform()

# Click verify
verify_btn = driver.find_element(By.ID, "recaptcha-verify-button")
verify_btn.click()

driver.switch_to.default_content()

titiritero

// Click each cell by index
const tableRows = await challengeFrame.$$('table.rc-imageselect-table tr');
for (const cellIdx of cells) {
  const row = Math.floor((cellIdx - 1) / (gridSize === '4x4' ? 4 : 3));
  const col = (cellIdx - 1) % (gridSize === '4x4' ? 4 : 3);
  const cell = (await tableRows[row].$$('td'))[col];
  await cell.click();
  await new Promise(r => setTimeout(r, 200));
}

await challengeFrame.click('#recaptcha-verify-button');

Manejo de mosaicos dinámicos

Algunas cuadrículas reCAPTCHA v2 reemplazan los mosaicos en los que se hace clic con nuevas imágenes. Maneje esto con un bucle de reintento:

def solve_with_dynamic_tiles(driver, api_key, max_rounds=3):
    for round_num in range(max_rounds):
        driver.switch_to.frame(challenge_iframe)

        # Re-capture grid and instruction
        img_b64 = driver.find_element(
            By.CSS_SELECTOR, "img.rc-image-tile-33"
        ).screenshot_as_base64

        # Submit and get cells (same as above)
        cells = submit_and_poll(api_key, img_b64, "3x3", keyword)

        if not cells:
            break

        # Click cells
        click_cells(driver, cells, "3x3")

        # Click verify
        driver.find_element(By.ID, "recaptcha-verify-button").click()

        driver.switch_to.default_content()
        time.sleep(2)

        # Check if solved (no more challenge iframe)
        try:
            driver.switch_to.frame(challenge_iframe)
            driver.switch_to.default_content()
        except Exception:
            return True  # Solved

    return False

Solución de problemas

Preguntas frecuentes

¿CaptchaAI admite cuadrículas de 4 × 4?

Sí. Configure grid_size=4x4 y la respuesta utilizará los índices 1-16.

¿Qué tan precisa es la resolución de imágenes de cuadrícula?

La precisión depende de la calidad de la imagen. Envíe la imagen CAPTCHA original sin recortar ni comprimir.

Resolver CAPTCHA de imágenes de cuadrícula con CaptchaAI

Obtenga su clave API encaptchaai.com.

Guías relacionadas

• Código de instrucciones BLS CAPTCHA
• Extracción de parámetros reCAPTCHA
• Preprocesamiento de imágenes CAPTCHA