haar小波变换的图像应用实例
Haar Wavelet 图片示例
import numpy as np
import cv2
import math
import requests
# numpy数组归一化
def Normalize(img):
_range = np.max(img) - np.min(img)
return (img - np.min(img)) / _range
#inputdata: row or column of image
#outputdata: array handle with haar_wavelet
def haar_dwt(row_or_col):#图片需要为2的次方形状
length=len(row_or_col)//2
Low_frequency=np.zeros(length,dtype=float)
High_frequency=np.zeros(length,dtype=float)
#小波变换的主体部分
for i in range(length):
Low_frequency[i]=(row_or_col[2*i]+row_or_col[2*i+1])/math.sqrt(2)
High_frequency[i]=(row_or_col[2*i]-row_or_col[2*i+1])/math.sqrt(2)
return np.append(Low_frequency,High_frequency)
#inputdata: array of image
#outputdata: array of image handel with haar_wavelet
def haar_dwt2D(img):
col_num=img.shape[1]
row_num=img.shape[0]
half_col_num=col_num//2
half_row_num=row_num//2
for i in range(row_num):
img[i]=haar_dwt(img[i])
for j in range(col_num):
img[:,j]=haar_dwt(img[:,j])
#为了强化图片的显示效果，对数据进行归一化处理
img[0:half_row_num,0:half_col_num]=Normalize(img[0:half_row_num,0:half_col_num])
img[half_row_num:row_num,0:half_col_num]=Normalize(img[half_row_num:row_num,0:half_col_num])
img[0:half_row_num,half_col_num:col_num]=Normalize(img[0:half_row_num,half_col_num:col_num])
img[half_row_num:row_num,half_col_num:col_num]=Normalize(img[half_row_num:row_num,half_col_num:col_num])
return img
if __name__ == '__main__':
#读取网络图片（2选1）
file_pic=requests.get('http://fanrencli.cn/fanrencli.cn/pic.png')
img= cv2.imdecode(np.fromstring(file_pic.content, np.uint8), 0).astype(np.float64)
#读取本地图片（2选1）
# img= cv2.imread("pic.png",0).astype(np.float64)
cv2.imshow('asd',haar_dwt2D(img))
cv2.waitKey(0)
for i in range(0,img.shape[0]//2-1,img.shape[0]//2-1):
for j in range(0,img.shape[1]//2-1,img.shape[1]//2-1):
img[i:i+img.shape[0]//2,j:j+img.shape[1]//2]=haar_dwt2D(img[i:i+img.shape[0]//2,j:j+img.shape[1]//2])
cv2.imshow('asd',img)
cv2.waitKey(0) 
