用python重写calcHist

在OpenCV中，calcHist函数用于计算图像的直方图。以下是使用Python重新实现calcHist函数的示例代码：

import numpy as np

def calcHist(images, channels, mask, histSize, ranges):
    """ Compute the histogram of the input image(s).

    :param images: input image(s) as a numpy array or list of numpy arrays.
    :param channels: list of indices of channels to be used for histogram calculation.
    :param mask: optional mask used to select pixels for histogram calculation.
    :param histSize: list of bin counts for each histogram dimension.
    :param ranges: list of ranges for each histogram dimension.

    :return: computed histogram(s) as a numpy array or list of numpy arrays.
    """
    # Check if images is a list of images or a single image
    if isinstance(images, list):
        # Compute histogram for each image in the list
        hist_list = []
        for image in images:
            hist = calcHist(image, channels, mask, histSize, ranges)
            hist_list.append(hist)
        return hist_list
    
    # Compute histogram for a single image
    hist = np.zeros(histSize, dtype=np.float32)
    dims = len(histSize)

    # Convert ranges to numpy arrays
    ranges = [np.asarray(r, dtype=np.float32) for r in ranges]

    # Compute the bin widths
    bin_widths = []
    for r, size in zip(ranges, histSize):
        bin_widths.append((r[1] - r[0]) / size)

    # Compute the histogram
    rows, cols = images.shape[:2]
    if dims == 1:
        for row in range(rows):
            for col in range(cols):
                if mask is None or mask[row, col]:
                    val = images[row, col, channels[0]]
                    if ranges[0][0] <= val < ranges[0][1]:
                        bin_index = int((val - ranges[0][0]) // bin_widths[0])
                        hist[bin_index] += 1
    elif dims == 2:
        for row in range(rows):
            for col in range(cols):
                if mask is None or mask[row, col]:
                    val1 = images[row, col, channels[0]]
                    val2 = images[row, col, channels[1]]
                    if ranges[0][0] <= val1 < ranges[0][1] and ranges[1][0] <= val2 < ranges[1][1]:
                        bin_index1 = int((val1 - ranges[0][0]) // bin_widths[0])
                        bin_index2 = int((val2 - ranges[1][0]) // bin_widths[1])
                        hist[bin_index1, bin_index2] += 1
    elif dims == 3:
        for row in range(rows):
            for col in range(cols):
                if mask is None or mask[row, col]:
                    val1 = images[row, col, channels[0]]
                    val2 = images[row, col, channels[1]]
                    val3 = images[row, col, channels[2]]
                    if ranges[0][0] <= val1 < ranges[0][1] and ranges[1][0] <= val2 < ranges[1][1] and ranges[2][0] <= val3 < ranges[2][1]:
                        bin_index1 = int((val1 - ranges[0][0]) // bin_widths[0])
                        bin_index2 = int((val2 - ranges[1][0]) // bin_widths[1])
                        bin_index3 = int((val3 - ranges[2][0]) // bin_widths[2])
                        hist[bin_index1, bin_index2, bin_index3] += 1
    else:
        raise ValueError("Images with more than 3 channels are not supported.")

    return hist

这个函数有五个输入参数和一个输出参数：

- images: 输入图像，可以是一个numpy数组或者一个numpy数组的列表
- channels: 用来计算直方图的通道，是一个索引列表
- mask: 一个可选的掩膜，只有掩膜内的像素才会被用来计算直方图
- histSize: 直方图的bin数目，是一个整数列表
- ranges: 每个bin的范围，是一个浮点数列表的列表

输出参数是一个numpy数组或者一个numpy数组的列表，代表计算出来的直方图。

这个函数的实现方式与OpenCV中的calcHist函数类似，但是在一些细节上有所不同。例如，这个函数使用了numpy数组的方法来遍历像素并计算直方图。另外，这个函数不支持多通道图像的直方图计算。