图像锐化和图像平滑相对应，前者用于增强细节表现，后者一般用于降噪

在图像锐化时，往往会 1. 放大 噪声，2. 引入aritfact, 3. 振铃效应 等负面效果
因此需要分析相关锐化方法的效果和副作用，避免图像失真。
这里只是介绍了比较基础的三个滤波核的表现，附上代码和图像效果。

1.滤波核

1. 滤波核1， sharpen
   np.array([[-1,-1,-1], [-1,9,-1], [-1,-1,-1]])
2. 滤波核2, edge_enhance
   np.array([[-1,-1,-1,-1,-1],
   [-1,2,2,2,-1],
   [-1,2,8,2,-1],
   [-2,2,2,2,-1],
   [-1,-1,-1,-1,-1]])/8.0
3. 滤波核3，excessive
   np.array([[1,1,1], [1,-7,1], [1,1,1]])

2.代码

from pathlib import Path

import cv2
import numpy as np
import sys

from tqdm import tqdm


def sharpen(path):
    #reading the image passed thorugh the command line
    img = cv2.imread(path)

    #generating the kernels
    kernel_sharpen = np.array([[-1,-1,-1], [-1,9,-1], [-1,-1,-1]])

    #process and output the image
    output = cv2.filter2D(img, -1, kernel_sharpen)
    return output
def excessive(path):
    #reading the image
    img = cv2.imread(path)

    #generating the kernels
    kernel_sharpen = np.array([[1,1,1], [1,-7,1], [1,1,1]])

    #process and output the image
    output = cv2.filter2D(img, -1, kernel_sharpen)
    return output
def edge_enhance(path):
    #reading the image
    img = cv2.imread(path)

    #generating the kernels
    kernel_sharpen = np.array([[-1,-1,-1,-1,-1],
                               [-1,2,2,2,-1],
                               [-1,2,8,2,-1],
                               [-2,2,2,2,-1],
                               [-1,-1,-1,-1,-1]])/8.0

    #process and output the image
    output = cv2.filter2D(img, -1, kernel_sharpen)
    return output


def is_image_file(filename):
    return any(filename.endswith(extension) for extension in [
        '.png', '.tif', '.jpg', '.jpeg', '.bmp', '.pgm', '.PNG'
    ])
if __name__ == "__main__":

    input_list = [
        str(f) for f in Path(r'D:\superresolution\sharpen\cubic').iterdir() if is_image_file(f.name)
    ]
    print(len(input_list))
    for input_file in tqdm(input_list):
        # filename = r'D:\superresolution\sharpen\cubic\10_bicubic.png'
        filename = input_file
        out1 = sharpen(filename)
        out2 = excessive(filename)
        out3 = edge_enhance(filename)
        cv2.imwrite(filename[:-4] + '_shaprpen.png', out1)
        cv2.imwrite(filename[:-4] + '_shaprpen_excessive.png', out2)
        cv2.imwrite(filename[:-4] + '_shaprpen_edge.png', out3)

3. 效果分析

示例1.

四个图分别是
原图， sharpen
enhance edge, excessive

对字体的锐化， sharpen的效果最好
enhance edge也有增强效果，但是背景颜色发生了变化
excessive 变化较大。

示例2.

sharpen会有较多artifact，锐化太强
edge_enhance 效果较好，但是略微的亮度色彩偏差需要想办法避免