import cv2
import numpy as np
import matplotlib.pyplot as plt

def mean_filter_builtin(image_path, kernel_size=3):
    """
    使用OpenCV内置函数实现均值滤波
    :param image_path: 输入图像路径
    :param kernel_size: 滤波器大小，必须为奇数
    :return: 原图和滤波后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    
    # 应用均值滤波
    filtered = cv2.blur(img, (kernel_size, kernel_size))
    
    return img, filtered

def manual_mean_filter(image_path, kernel_size=3):
    """
    手动实现均值滤波
    :param image_path: 输入图像路径
    :param kernel_size: 滤波器大小，必须为奇数
    :return: 均值滤波后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    img_float = img.astype(np.float32)
    height, width, channels = img.shape
    
    # 计算边界填充大小
    pad_size = kernel_size // 2
    
    # 对图像进行边界填充
    padded_img = np.pad(img_float, ((pad_size, pad_size), (pad_size, pad_size), (0, 0)), mode='edge')
    
    # 创建输出图像
    filtered = np.zeros_like(img_float)
    
    # 应用均值滤波
    for i in range(height):
        for j in range(width):
            for c in range(channels):
                # 提取当前窗口
                window = padded_img[i:i+kernel_size, j:j+kernel_size, c]
                # 计算平均值
                filtered[i, j, c] = np.mean(window)
    
    return img, filtered.astype(np.uint8)

def separable_mean_filter(image_path, kernel_size=3):
    """
    使用可分离滤波器实现均值滤波（提高效率）
    :param image_path: 输入图像路径
    :param kernel_size: 滤波器大小，必须为奇数
    :return: 原图和滤波后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    
    # 创建一维均值滤波器
    kernel_1d = np.ones(kernel_size) / kernel_size
    
    # 分别在水平和垂直方向应用滤波器
    temp = cv2.filter2D(img, -1, kernel_1d.reshape(1, -1))  # 水平方向
    filtered = cv2.filter2D(temp, -1, kernel_1d.reshape(-1, 1))  # 垂直方向
    
    return img, filtered

def adaptive_mean_filter(image_path, noise_variance, kernel_size=3):
    """
    自适应均值滤波
    :param image_path: 输入图像路径
    :param noise_variance: 噪声方差估计
    :param kernel_size: 滤波器大小
    :return: 原图和自适应滤波后的图像
    """
    # 读取图像
    img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE).astype(np.float32)
    height, width = img.shape
    
    # 计算边界填充大小
    pad_size = kernel_size // 2
    
    # 对图像进行边界填充
    padded_img = np.pad(img, pad_size, mode='edge')
    
    # 创建输出图像
    filtered = np.zeros_like(img)
    
    # 计算局部统计信息
    for i in range(height):
        for j in range(width):
            # 提取当前窗口
            window = padded_img[i:i+kernel_size, j:j+kernel_size]
            
            # 计算局部均值和方差
            local_mean = np.mean(window)
            local_variance = np.var(window)
            
            # 如果局部方差小于噪声方差，则认为该区域是平坦区域
            # 否则应用自适应滤波
            if local_variance == 0:
                filtered[i, j] = img[i, j]
            else:
                weight = min(noise_variance / local_variance, 1)
                filtered[i, j] = img[i, j] - weight * (img[i, j] - local_mean)
    
    return cv2.cvtColor(img.astype(np.uint8), cv2.COLOR_GRAY2BGR), cv2.cvtColor(filtered.astype(np.uint8), cv2.COLOR_GRAY2BGR)

def weighted_mean_filter(image_path, weights=None):
    """
    加权均值滤波
    :param image_path: 输入图像路径
    :param weights: 自定义权重矩阵
    :return: 原图和加权滤波后的图像
    """
    if weights is None:
        # 默认使用高斯权重
        weights = np.array([
            [1, 2, 1],
            [2, 4, 2],
            [1, 2, 1]
        ], dtype=np.float32)
        weights /= np.sum(weights)
    
    # 读取图像
    img = cv2.imread(image_path)
    
    # 应用加权均值滤波
    filtered = cv2.filter2D(img, -1, weights)
    
    return img, filtered

# 使用示例
if __name__ == "__main__":
    # 注意：需要提供实际的图像路径
    # img, result = mean_filter_builtin('image.jpg', 5)
    # manual_img, manual_result = manual_mean_filter('image.jpg', 5)
    # sep_img, sep_result = separable_mean_filter('image.jpg', 5)
    # adapt_img, adapt_result = adaptive_mean_filter('image.jpg', 10, 5)
    # weight_img, weight_result = weighted_mean_filter('image.jpg')
    pass