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

def bilateral_filter_builtin(image_path, d=15, sigma_color=80, sigma_space=80):
    """
    使用OpenCV内置函数实现双边滤波
    :param image_path: 输入图像路径
    :param d: 邻域直径
    :param sigma_color: 颜色空间的标准差
    :param sigma_space: 坐标空间的标准差
    :return: 原图和滤波后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    
    # 应用双边滤波
    filtered = cv2.bilateralFilter(img, d, sigma_color, sigma_space)
    
    return img, filtered

def manual_bilateral_filter(image_path, d=15, sigma_color=80, sigma_space=80):
    """
    手动实现双边滤波
    :param image_path: 输入图像路径
    :param d: 邻域直径
    :param sigma_color: 颜色空间的标准差
    :param sigma_space: 坐标空间的标准差
    :return: 双边滤波后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    img_float = img.astype(np.float64)
    height, width, channels = img.shape
    
    # 计算边界填充大小
    pad_size = d // 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)
    
    # 计算空间权重
    space_coeff = -0.5 / (sigma_space * sigma_space)
    color_coeff = -0.5 / (sigma_color * sigma_color)
    
    # 应用双边滤波
    for i in range(height):
        for j in range(width):
            # 计算当前像素的权重中心
            center_pixel = padded_img[i + pad_size, j + pad_size]
            
            # 初始化权重和加权总和
            weight_sum = np.zeros(channels)
            weighted_sum = np.zeros((channels,))
            
            # 遍历邻域
            for ki in range(-pad_size, pad_size + 1):
                for kj in range(-pad_size, pad_size + 1):
                    neighbor_pixel = padded_img[i + pad_size + ki, j + pad_size + kj]
                    
                    # 计算空间距离权重
                    spatial_dist_sq = ki * ki + kj * kj
                    spatial_weight = np.exp(spatial_dist_sq * space_coeff)
                    
                    # 计算颜色距离权重
                    color_dist_sq = np.sum((center_pixel - neighbor_pixel) ** 2)
                    color_weight = np.exp(color_dist_sq * color_coeff)
                    
                    # 总权重
                    total_weight = spatial_weight * color_weight
                    
                    # 累加权重和加权像素值
                    weight_sum += total_weight
                    weighted_sum += neighbor_pixel * total_weight
            
            # 计算加权平均
            filtered[i, j] = weighted_sum / weight_sum
    
    return img, filtered.astype(np.uint8)

def edge_preserving_smoothing(image_path, d=15, sigma_color=80, sigma_space=80):
    """
    边缘保持平滑（双边滤波的应用）
    :param image_path: 输入图像路径
    :param d: 邻域直径
    :param sigma_color: 颜色空间的标准差
    :param sigma_space: 坐标空间的标准差
    :return: 原图和边缘保持平滑后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    
    # 应用双边滤波以保持边缘的同时平滑图像
    smoothed = cv2.bilateralFilter(img, d, sigma_color, sigma_space)
    
    return img, smoothed

def detail_enhancement(image_path, d=9, sigma_color=75, sigma_space=75):
    """
    细节增强（通过减去平滑图像）
    :param image_path: 输入图像路径
    :param d: 邻域直径
    :param sigma_color: 颜色空间的标准差
    :param sigma_space: 坐标空间的标准差
    :return: 原图和细节增强后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    img_float = img.astype(np.float64)
    
    # 应用双边滤波获得平滑版本
    smooth = cv2.bilateralFilter(img, d, sigma_color, sigma_space)
    
    # 计算细节（原图减去平滑图）
    detail = img_float - smooth.astype(np.float64)
    
    # 增强细节（可调整增强系数）
    enhanced = img_float + 0.5 * detail
    
    # 确保像素值在有效范围内
    enhanced = np.clip(enhanced, 0, 255)
    
    return img, enhanced.astype(np.uint8)

def texture_filtering(image_path, d=20, sigma_color=50, sigma_space=50):
    """
    纹理过滤（保留主要结构，去除细节纹理）
    :param image_path: 输入图像路径
    :param d: 邻域直径
    :param sigma_color: 颜色空间的标准差
    :param sigma_space: 坐标空间的标准差
    :return: 原图和纹理过滤后的图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    
    # 使用较大的参数来过滤纹理
    filtered = cv2.bilateralFilter(img, d, sigma_color, sigma_space)
    
    return img, filtered

def cartoon_effect(image_path):
    """
    卡通效果（结合双边滤波和边缘检测）
    :param image_path: 输入图像路径
    :return: 原图和卡通效果图像
    """
    # 读取图像
    img = cv2.imread(image_path)
    
    # 步骤1：多次应用双边滤波以获得平滑效果
    filtered = img
    for _ in range(2):
        filtered = cv2.bilateralFilter(filtered, 9, 200, 200)
    
    # 步骤2：转换为灰度图并应用边缘检测
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    edges = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 9, 9)
    
    # 步骤3：将边缘图转换为彩色
    edges_colored = cv2.cvtColor(edges, cv2.COLOR_GRAY2BGR)
    
    # 步骤4：将边缘图与平滑图结合
    cartoon = cv2.bitwise_and(filtered, edges_colored)
    
    return img, cartoon

# 使用示例
if __name__ == "__main__":
    # 注意：需要提供实际的图像路径
    # img, result = bilateral_filter_builtin('image.jpg', 15, 80, 80)
    # manual_img, manual_result = manual_bilateral_filter('image.jpg', 15, 80, 80)
    # smooth_img, smooth_result = edge_preserving_smoothing('image.jpg', 15, 80, 80)
    # enhance_img, enhance_result = detail_enhancement('image.jpg', 9, 75, 75)
    # texture_img, texture_result = texture_filtering('image.jpg', 20, 50, 50)
    # cartoon_img, cartoon_result = cartoon_effect('image.jpg')
    pass