线条的聚类/抑制（NMS）

线条的聚类和抑制（Non-Maximum Suppression, NMS）是一种常用的计算机视觉算法，用于检测图像中的边缘或线条。下面是一个使用Python和OpenCV库实现线条聚类和抑制的示例代码：

import cv2
import numpy as np

def line_clustering(lines, threshold):
    clusters = []
    
    for line in lines:
        found_cluster = False
        
        for cluster in clusters:
            if abs(line[0] - cluster[0]) < threshold and abs(line[1] - cluster[1]) < threshold:
                cluster.append(line)
                found_cluster = True
                break
        
        if not found_cluster:
            clusters.append([line])
    
    return clusters

def nms(lines, overlap_threshold):
    lines = sorted(lines, key=lambda x: x[2], reverse=True)
    selected_lines = []
    
    while len(lines) > 0:
        best_line = lines[0]
        selected_lines.append(best_line)
        lines = lines[1:]
        
        remaining_lines = []
        for line in lines:
            overlap = min(best_line[3], line[3]) - max(best_line[2], line[2])
            if overlap <= 0:
                remaining_lines.append(line)
        
        lines = remaining_lines
    
    return selected_lines

# 读取图像并提取边缘
image = cv2.imread('image.jpg', 0)
edges = cv2.Canny(image, 50, 150)

# 提取直线
lines = cv2.HoughLinesP(edges, 1, np.pi/180, 100, minLineLength=100, maxLineGap=10)

# 进行线条聚类
threshold = 10
clusters = line_clustering(lines, threshold)

# 画出聚类后的线条
clustered_image = np.zeros_like(image)
for cluster in clusters:
    color = np.random.randint(0, 255, 3).tolist()
    for line in cluster:
        x1, y1, x2, y2 = line[0]
        cv2.line(clustered_image, (x1, y1), (x2, y2), color, 2)

# 进行非极大值抑制
overlap_threshold = 0.3
selected_lines = nms(lines, overlap_threshold)

# 画出抑制后的线条
nms_image = np.zeros_like(image)
for line in selected_lines:
    x1, y1, x2, y2 = line[0]
    cv2.line(nms_image, (x1, y1), (x2, y2), (255, 255, 255), 2)

# 显示图像
cv2.imshow('Original Image', image)
cv2.imshow('Edge Image', edges)
cv2.imshow('Clustered Image', clustered_image)
cv2.imshow('NMS Image', nms_image)
cv2.waitKey(0)
cv2.destroyAllWindows()

请注意，以上代码仅为示例，实际应用中可能需要根据具体情况进行调整和优化。