#!/usr/bin/env python # -*- coding: utf-8 -*- """ 临时测试脚本：对 pic1.png 与 Screenshot-pic1.png 做多组参数匹配，直到找到匹配并在截图上标出。 用法: 在项目根目录执行 python python/scripts/test_match_pic1_screenshot.py 或 cd static/process/GenerateNote && python ../../../python/scripts/test_match_pic1_screenshot.py 输出: 控制台打印最佳参数与坐标，并生成 Screenshot-pic1-marked.png 标出匹配框与中心点。 """ from __future__ import print_function import os import sys import json # 项目根目录（本脚本在 python/scripts/） SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__)) PROJECT_ROOT = os.path.normpath(os.path.join(SCRIPT_DIR, '..', '..')) if PROJECT_ROOT not in sys.path: sys.path.insert(0, PROJECT_ROOT) # 默认路径：流程目录下的 tmp PROCESS_DIR = os.path.join(PROJECT_ROOT, 'static', 'process', 'GenerateNote') TMP_DIR = os.path.join(PROCESS_DIR, 'tmp') SCREENSHOT_PATH = os.path.join(TMP_DIR, 'Screenshot-pic1.png') TEMPLATE_PATH = os.path.join(TMP_DIR, 'pic1.png') MARKED_OUT_PATH = os.path.join(TMP_DIR, 'Screenshot-pic1-marked.png') try: import cv2 import numpy as np except ImportError: print('请安装: pip install opencv-python numpy') sys.exit(1) # 动态加载 image-match 模块（文件名带横杠） import importlib.util _image_match_path = os.path.join(SCRIPT_DIR, 'image-match.py') _spec = importlib.util.spec_from_file_location('image_match', _image_match_path) _image_match = importlib.util.module_from_spec(_spec) _spec.loader.exec_module(_image_match) load_image = _image_match.load_image multi_scale_template_match = _image_match.multi_scale_template_match match_by_features = _image_match.match_by_features HAS_ROMA = getattr(_image_match, 'HAS_ROMA', False) match_by_roma = getattr(_image_match, 'match_by_roma', lambda *a, **k: None) def multi_scale_with_correlation(screenshot, template, threshold=0.50, scale_min=0.4, scale_max=1.65, steps=80): """多尺度模板匹配并返回最佳相关系数。返回 (x, y, w, h, center_x, center_y, best_val) 或 None。""" gray_screen = cv2.cvtColor(screenshot, cv2.COLOR_BGR2GRAY) gray_tpl = cv2.cvtColor(template, cv2.COLOR_BGR2GRAY) sh, sw = screenshot.shape[:2] t_h, t_w = template.shape[:2] best = None best_val = threshold step = max(0.02, (scale_max - scale_min) / float(steps)) for scale in np.arange(scale_min, scale_max + step * 0.5, step): w = max(8, int(round(t_w * scale))) h = max(8, int(round(t_h * scale))) if h > sh or w > sw: continue resized = cv2.resize(gray_tpl, (w, h), interpolation=cv2.INTER_AREA) result = cv2.matchTemplate(gray_screen, resized, cv2.TM_CCOEFF_NORMED) min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(result) if max_val > best_val: best_val = max_val x, y = int(max_loc[0]), int(max_loc[1]) center_x = x + w // 2 center_y = y + h // 2 best = (x, y, w, h, center_x, center_y, best_val) if best is None: return None return best def draw_match(screenshot, x, y, w, h, center_x, center_y, out_path): """在截图上画匹配框和中心点并保存。""" img = screenshot.copy() cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2) cv2.circle(img, (center_x, center_y), 8, (0, 0, 255), 2) cv2.putText(img, 'center', (center_x + 10, center_y), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 255), 1) cv2.imwrite(out_path, img) print('已保存标注图:', out_path) def main(): screenshot_path = SCREENSHOT_PATH template_path = TEMPLATE_PATH if len(sys.argv) >= 3: screenshot_path = sys.argv[1] template_path = sys.argv[2] if not os.path.isfile(screenshot_path): print('截图不存在:', screenshot_path) sys.exit(1) if not os.path.isfile(template_path): print('模板不存在:', template_path) sys.exit(1) screenshot = load_image(screenshot_path) template = load_image(template_path) if screenshot is None: print('无法加载截图') sys.exit(1) if template is None: print('无法加载模板') sys.exit(1) sh, sw = screenshot.shape[:2] t_h, t_w = template.shape[:2] print('截图尺寸: {}x{}, 模板尺寸: {}x{}'.format(sw, sh, t_w, t_h)) # 1) 尝试 RoMa（若可用），仅当中心在上半区时才采用 if HAS_ROMA: roma_result = match_by_roma(screenshot, template, min_matches=4) if roma_result is not None: x, y, w, h, center_x, center_y = roma_result if center_y <= sh * 0.6: print('[RoMa] 匹配成功: center=({}, {})'.format(center_x, center_y)) draw_match(screenshot, x, y, w, h, center_x, center_y, MARKED_OUT_PATH) print('推荐 process.json: 保持 [0.15, 2.0], 方形裁剪如 [0.8, "w"]') sys.exit(0) # 2) 收集多尺度模板匹配结果（参数网格），优先取「中心在画面上半区」的匹配（格子图通常在上方） sh, sw = screenshot.shape[:2] upper_ratio = 0.6 # 中心 y < sh * upper_ratio 视为上半区（格子区域） candidates = [] # 3) 多尺度模板匹配：参数网格（缩略图通常 scale 约 0.1～0.4） scale_min_list = [0.05, 0.10, 0.15] scale_max_list = [1.0, 1.5, 2.0] threshold_list = [0.40, 0.46, 0.52] best_result = None best_corr = 0.0 best_params = None for scale_min in scale_min_list: for scale_max in scale_max_list: if scale_max <= scale_min: continue for th in threshold_list: r = multi_scale_with_correlation( screenshot, template, threshold=th, scale_min=scale_min, scale_max=scale_max, steps=50 ) if r is not None: x, y, w, h, cx, cy, corr = r candidates.append((corr, (x, y, w, h, cx, cy), (scale_min, scale_max, th))) # 优先选中心在上半区的最高相关匹配，否则选全局最高相关 best_result = None best_corr = 0.0 best_params = None for corr, rect, params in sorted(candidates, key=lambda t: -t[0]): x, y, w, h, cx, cy = rect if cy <= sh * upper_ratio and corr > best_corr: best_corr = corr best_result = rect best_params = params if best_result is None and candidates: best_corr, best_result, best_params = max(candidates, key=lambda t: t[0]) print('(未在上半区找到匹配，使用全局最高相关；请查看标注图确认是否为底部缩略图)') if best_result is None: print('未找到任何匹配。可尝试: 降低 threshold、扩大 scale_min~scale_max、或检查模板是否在截图内。') sys.exit(1) x, y, w, h, center_x, center_y = best_result scale_min, scale_max, th = best_params print('') print('[多尺度模板] 匹配成功 (最佳相关系数 {:.3f}, 中心在上半区)'.format(best_corr)) print(' 中心坐标: ({}, {})'.format(center_x, center_y)) print(' 矩形: x={} y={} w={} h={}'.format(x, y, w, h)) print(' 推荐 process.json inVars:') print(' inVars: ["tmp/pic{{idx}}.png", [{}, {}], [0.8, "w"]]'.format(scale_min, scale_max)) print(' 并在 image-match.py 中 fallback 阈值可设为 <= {:.2f}'.format(th)) print('') draw_match(screenshot, x, y, w, h, center_x, center_y, MARKED_OUT_PATH) if __name__ == '__main__': main()