自己用python代码编写函数实现自适应直方图均衡化，且生成图比原图小一圈

好的，我已经明白了您的需求，下面是使用 Python 实现自适应直方图均衡化并生成比原图小一圈的函数：

import cv2
import numpy as np

def adaptive_histogram_equalization(img):
    # 将图像转为灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    
    # 计算图像的均值和方差
    mean, std = cv2.meanStdDev(gray)
    mean = mean[0][0]
    std = std[0][0]
    
    # 设置参数
    k = 0.4
    epsilon = 0.0001
    max_intensity = 255
    
    # 计算每个像素的局部均值和局部标准差
    width, height = gray.shape
    local_mean = np.zeros((width, height))
    local_std = np.zeros((width, height))
    for i in range(width):
        for j in range(height):
            x0 = max(0, i - 1)
            y0 = max(0, j - 1)
            x1 = min(width - 1, i + 1)
            y1 = min(height - 1, j + 1)
            local_patch = gray[x0:x1, y0:y1]
            local_mean[i][j] = np.mean(local_patch)
            local_std[i][j] = np.std(local_patch)

    # 计算增益和偏移量
    gain = k * std / (local_std + epsilon)
    offset = mean - gain * local_mean
    
    # 进行自适应直方图均衡化
    result = np.zeros_like(gray)
    for i in range(width):
        for j in range(height):
            value = int(gain[i][j] * gray[i][j] + offset[i][j])
            value = min(max_intensity, max(0, value))
            result[i][j] = value
    
    # 将结果进行缩小
    result = cv2.resize(result, (result.shape[1]-2, result.shape[0]-2))
    
    return result

这个函数的输入为一张彩色图片，输出为经过自适应直方图均衡化并缩小一圈后的灰度图像。您可以将图片读入后传入这个函数，例如：

img = cv2.imread('example.jpg')
result = adaptive_histogram_equalization(img)
cv2.imshow('result', result)
cv2.waitKey(0)
cv2.destroyAllWindows()

其中，`example.jpg` 为您要处理的图片文件名。