utils/iqa_recognize.py

import numpy as np
import cv2
import torch
from PIL import Image
import termcolor
from glob import glob


template_dir = "character_template"
char_info = {
    "character_template/e.png": "鄂", "character_template/gui.png": "桂",
    "character_template/hei.png": "黑", "character_template/ji.png": "冀",
    "character_template/gui1.png": "贵", "character_template/jing.png": "京",
    "character_template/lu.png": "鲁", "character_template/min.png": "闽",
    "character_template/su.png": "苏", "character_template/wan.png": "皖",
    "character_template/yu.png": "豫", "character_template/yue.png": "粤",
    "character_template/xin.png": "新", "character_template/chuan.jpg": "川",
    "character_template/ji1.jpg": "吉", "character_template/jin.jpg": "津",
    "character_template/liao.jpg": "辽", "character_template/shan.jpg": "陕",
    "character_template/zhe.jpg": "浙", "character_template/meng.jpg": "蒙",
}
char_list = list(char_info.values())
character_image_list = []
for template_path in char_info.keys():
    character_image = Image.open(template_path).convert('RGB')
    character_image_list.append(character_image)
print(f"Support Chinese characters: {termcolor.colored(char_list, 'blue')}")


def calculate_correlation(image1: Image.Image, image2: Image.Image):
    image1 = np.array(image1)
    image2 = np.array(image2)

    image1 = cv2.cvtColor(image1, cv2.COLOR_BGR2GRAY)
    image2 = cv2.cvtColor(image2, cv2.COLOR_BGR2GRAY)
    image2 = cv2.resize(image2, (image1.shape[1], image1.shape[0]))
    
    image1_flat = image1.flatten()
    image2_flat = image2.flatten()
    
    correlation = np.corrcoef(image1_flat, image2_flat)[0, 1]
    return correlation


def calculate_sift(image1: Image.Image, image2: Image.Image):
    image1 = np.array(image1)
    image2 = np.array(image2)

    image1 = cv2.cvtColor(image1, cv2.COLOR_BGR2GRAY)
    image2 = cv2.cvtColor(image2, cv2.COLOR_BGR2GRAY)
    image2 = cv2.resize(image2, (image1.shape[1], image1.shape[0]))

    sift = cv2.SIFT_create()
    kp1, des1 = sift.detectAndCompute(image1, None)
    kp2, des2 = sift.detectAndCompute(image2, None)

    bf = cv2.BFMatcher()
    matches = bf.knnMatch(des1, des2, k=2)
    
    good = []
    for m, n in matches:
        if m.distance < 0.75 * n.distance:
            good.append(m)

    src_pts = np.float32([kp1[m.queryIdx].pt for m in good]).reshape(-1, 1, 2)
    dst_pts = np.float32([kp2[m.trainIdx].pt for m in good]).reshape(-1, 1, 2)

    if len(good) < 4:
        return len(good)

    homography, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0)

    inlier_matches = [m for i, m in enumerate(good) if mask[i] == 1]

    return len(inlier_matches)


def recognize_chinese_char(image: Image.Image, image_path: str=None, print_probs=False):
    if image_path is not None:
        image = Image.open(image_path).convert('RGB')


    score_list = []
    for character_image in character_image_list:
        score_list.append(calculate_sift(image, character_image))
    char_index = np.array(score_list).argmax()

    if print_probs:
        prob_dict = dict(zip(char_list, score_list))
        print(f"Label probs: {termcolor.colored(prob_dict, 'red')}")
    return char_list[char_index]


if __name__ == "__main__":
    img_paths = glob(f"cut_plate/*.jpg") + glob(f"cut_plate/*.png") + glob(f"cut_plate/*.jpeg")
    for image_path in img_paths:
        print(image_path, recognize_chinese_char(None, image_path))