记录反光图像二值化，滤波，centertiqu

已于 2025-05-14 08:03:42 修改
阅读量324 收藏 10
点赞数 5
CC 4.0 BY-SA版权
文章标签： python
于 2025-05-13 11:39:27 首次发布
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.youkuaiyun.com/zzgiack/article/details/147920119 
import cv2
import numpy as np
import os
import glob

def extract_laser_center2(img: np.ndarray, threshold: int = 100) -> np.ndarray:    
    if img.ndim == 3 and img.shape[2] == 3:
        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    else:
        gray = img.copy()

    rows, cols = gray.shape
    center_map = np.zeros_like(gray, dtype=np.uint8)

    for col in range(cols):
        col_vals = gray[:, col].astype(np.int32)
        mask = col_vals > threshold
        if np.any(mask):
            weights = col_vals[mask]
            ys = np.nonzero(mask)[0]
            y_center = (weights * ys).sum() / weights.sum()
            x = col
            y = int(round(y_center))
            if 0 <= y < rows:
                center_map[y, x] = 255

    return center_map

# ------------------- 配置参数 -------------------
input_folder = r"G:\1126_1106\0509\images_20250509_133744"
image_extensions = ['*.jpg', '*.png', '*.bmp']
save_results = False  # 如需保存处理图像，设为 True
output_folder = os.path.join(input_folder, "processed_results")

if save_results and not os.path.exists(output_folder):
    os.makedirs(output_folder)

# 扫描所有图像文件
image_paths = []
for ext in image_extensions:
    image_paths.extend(glob.glob(os.path.join(input_folder, ext)))

print(f"检测到 {len(image_paths)} 张图像进行批量处理...")

for idx, img_path in enumerate(image_paths, 1):
    print(f"\n[{idx}/{len(image_paths)}] 正在处理：{os.path.basename(img_path)}")

    image = cv2.imread(img_path, cv2.IMREAD_GRAYSCALE)
    if image is None:
        print(f"⚠️  无法读取：{img_path}")
        continue

    # ----- 亮度与对比度调整 -----
    contrast = 110 / 100.0
    brightness = -13
    adjusted = image.astype(np.float32) * contrast + brightness
    adjusted = np.clip(adjusted, 0, 255).astype(np.uint8)

    # ----- Gamma校正 -----
    gamma = 10 / 100.0
    invGamma = 1.0 / gamma
    table = np.array([((i / 255.0) ** invGamma) * 255 for i in np.arange(256)]).astype("uint8")
    adjusted_gamma = cv2.LUT(adjusted, table)

    # ----- 中值滤波 -----
    filtered = cv2.medianBlur(adjusted_gamma, 11)

    # ----- 高斯滤波 -----
    blurred = cv2.GaussianBlur(filtered, (5, 7), 0)

    # ----- Otsu 二值化 -----
    _, otsu_thresh = cv2.threshold(blurred, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)

    # ----- 形态学清理 -----
    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 3))
    binary_cleaned = cv2.morphologyEx(otsu_thresh, cv2.MORPH_CLOSE, kernel)
    binary_cleaned = cv2.morphologyEx(binary_cleaned, cv2.MORPH_OPEN, kernel)

    # ----- 连通域分析 + 面积筛选 -----
    num_labels, labels, stats, centroids = cv2.connectedComponentsWithStats(binary_cleaned, connectivity=8)
    filtered_components = np.zeros_like(binary_cleaned)
    for i in range(1, num_labels):
        if stats[i, cv2.CC_STAT_AREA] >= 100:
            filtered_components[labels == i] = 255

    # ----- 激光中心提取 -----
    laser_center = extract_laser_center2(filtered_components, threshold=100)

    # ----- 显示结果 -----
    cv2.imshow("原图", image)
    cv2.imshow("激光中心", laser_center)
    cv2.imshow("连通域筛选", filtered_components)
    key = cv2.waitKey()  # 可以调整延时

    # ----- 可选：保存结果 -----
    if save_results:
        base_name = os.path.splitext(os.path.basename(img_path))[0]
        cv2.imwrite(os.path.join(output_folder, f"{base_name}_center.png"), laser_center)
        cv2.imwrite(os.path.join(output_folder, f"{base_name}_binary.png"), filtered_components)

cv2.destroyAllWindows()
print("\n✅ 批量处理完成！")

继续针对金属反光进行优化：


import cv2
import numpy as np
import os
import glob

def extract_laser_center2(img: np.ndarray, threshold: int = 100) -> np.ndarray:    
    if img.ndim == 3 and img.shape[2] == 3:
        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    else:
        gray = img.copy()

    rows, cols = gray.shape
    center_map = np.zeros_like(gray, dtype=np.uint8)

    for col in range(cols):
        col_vals = gray[:, col].astype(np.int32)
        mask = col_vals > threshold
        if np.any(mask):
            weights = col_vals[mask]
            ys = np.nonzero(mask)[0]
            y_center = (weights * ys).sum() / weights.sum()
            x = col
            y = int(round(y_center))
            if 0 <= y < rows:
                center_map[y, x] = 255

    return center_map

# ------------------- 配置参数 -------------------
input_folder = r"G:\1126_1106\0509\images_20250509_103731"
image_extensions = ['*.jpg', '*.png', '*.bmp']
save_results = False  # 如需保存处理图像，设为 True
output_folder = os.path.join(input_folder, "processed_results")

if save_results and not os.path.exists(output_folder):
    os.makedirs(output_folder)

# 扫描所有图像文件
image_paths = []
for ext in image_extensions:
    image_paths.extend(glob.glob(os.path.join(input_folder, ext)))

print(f"检测到 {len(image_paths)} 张图像进行批量处理...")

for idx, img_path in enumerate(image_paths, 1):
    print(f"\n[{idx}/{len(image_paths)}] 正在处理：{os.path.basename(img_path)}")

    image = cv2.imread(img_path, cv2.IMREAD_GRAYSCALE)
    if image is None:
        print(f"⚠️  无法读取：{img_path}")
        continue

    # ----- 亮度与对比度调整 -----
    contrast = 110 / 100.0
    brightness = -13
    adjusted = image.astype(np.float32) * contrast + brightness
    adjusted = np.clip(adjusted, 0, 255).astype(np.uint8)

    # ----- Gamma校正 -----
    gamma = 10 / 100.0
    invGamma = 1.0 / gamma
    table = np.array([((i / 255.0) ** invGamma) * 255 for i in np.arange(256)]).astype("uint8")
    adjusted_gamma = cv2.LUT(adjusted, table)
    
    d=50
    sigmaColor=140
    sigmaSpace=150
    dst = cv2.bilateralFilter(adjusted_gamma, d, sigmaColor, sigmaSpace)
    # ----- 中值滤波 -----
    filtered = cv2.medianBlur(dst, 5)

    # ----- 高斯滤波 -----
    #blurred = cv2.GaussianBlur(filtered, (3, 3), 0)
    blurred = filtered
    # ----- Otsu 二值化 -----
    _, otsu_thresh = cv2.threshold(blurred, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)

    # ----- 形态学清理 -----
    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 3))
    binary_cleaned = cv2.morphologyEx(otsu_thresh, cv2.MORPH_CLOSE, kernel)
    binary_cleaned = cv2.morphologyEx(binary_cleaned, cv2.MORPH_OPEN, kernel)

    # ----- 连通域分析 + 面积筛选 -----
    num_labels, labels, stats, centroids = cv2.connectedComponentsWithStats(binary_cleaned, connectivity=8)
    filtered_components = np.zeros_like(binary_cleaned)
    for i in range(1, num_labels):
        if stats[i, cv2.CC_STAT_AREA] >= 50:
            filtered_components[labels == i] = 255

    # ----- 激光中心提取 -----
    laser_center = extract_laser_center2(filtered_components, threshold=100)

    # ----- 显示结果 -----
    cv2.imshow("img", image)
    #cv2.imshow("bilateralFilter", dst)
    cv2.imshow("laser_center", laser_center)
    #cv2.imshow("filtered_components", filtered_components)
    key = cv2.waitKey()  # 可以调整延时

    # ----- 可选：保存结果 -----
    if save_results:
        base_name = os.path.splitext(os.path.basename(img_path))[0]
        cv2.imwrite(os.path.join(output_folder, f"{base_name}_center.png"), laser_center)
        cv2.imwrite(os.path.join(output_folder, f"{base_name}_binary.png"), filtered_components)

cv2.destroyAllWindows()
print("\n✅ 批量处理完成！")
George_zhai97
关注
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值