基于python的人群密度图生成

最新推荐文章于 2025-03-26 16:55:15 发布

原创

最新推荐文章于 2025-03-26 16:55:15 发布 · 5.9k 阅读

40 ·

CC 4.0 BY-SA版权

文章标签：

#人群密度图生成

本文介绍了一种基于高斯滤波快速实现人群密度图的方法，包括如何计算高斯核参数、预处理图像数据及使用KD树进行最近邻搜索等关键技术细节。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

相关原理见：https://zhuanlan.zhihu.com/p/39424587

相关代码如下：

import os
import cv2
import glob
import h5py
import scipy
import pickle
import numpy as np
from PIL import Image
import scipy.io as io
from itertools import islice
from tqdm import tqdm
from matplotlib import pyplot as plt
from sortedcontainers import SortedDict
from scipy.ndimage.filters import gaussian_filter


def get_img_pathes(path_sets):
    """
    Return all images from all pathes in 'path_sets'
    """
    img_pathes = []
    for path in path_sets:
        for img_path in glob.glob(os.path.join(path, '*.jpg')):
            img_pathes.append(img_path)
    return img_pathes


def save_computed_density(density_map, out_path):
    """
    Save density map to h5py format
    """
    with h5py.File(out_path, 'w') as hf:
        hf['density'] = density_map


def compute_sigma(gt_count, distance=None, min_sigma=1, method=1, fixed_sigma=15):
    """
    Compute sigma for gaussian kernel with different methods :
    * method = 1 : sigma = (sum of distance to 3 nearest neighbors) / 10
    * method = 2 : sigma = distance to nearest neighbor
    * method = 3 : sigma = fixed value
    ** if sigma lower than threshold 'min_sigma', then 'min_sigma' will be used
    ** in case of one poin