图像融合

##拉普拉斯金字塔

In CollageCreate function, there is a Fusion class object. The Fusion class will get the image folder name and contains different kinds of methods to process the images. You can put different numbers of images into the folder and it will process all these images and output a fusion image. In the class , it will first load the images and then calculate all their Laplacian pyramids. I use the pyramids to fuse the images in multiscale layers. Finally , I upsample the fused Laplacian pyramids and use all of them to create a new fusion type. So far, the fusion image is created and I plot this image in the end.

import numpy as np
import matplotlib.pyplot as plt
from skimage import data, transform, color,exposure, img_as_float, io
import osclass Fusion(object):def __init__(self,AddressofFolder):self.folder_name = AddressofFolderself.n_images = Noneself.images = self.load_images(self.folder_name)self.pyramids = self.get_laplacian()self.ls = Noneself.stack_ls()def load_images(self,path):names = os.listdir(path)self.n_images = len(names)images = [io.imread(os.path.join('images',n)) for n in names]new_ = []for i in images:new_.append(transform.resize(i,(600,600,3)))return new_def get_laplacian(self):ls = []for i in self.images:lp = list(transform.pyramid_laplacian(i, downscale=2, multichannel=True))ls.append(lp)return lsdef stack_ls(self):layers = len(self.pyramids[0])n1 = self.n_images//2n2 = self.n_images - n1ls1_imgs = []for l in range(0,layers):temp = self.pyramids[0][l]for i in range(1,n1):temp = np.hstack((temp[:,:],self.pyramids[i][l][:,:]))ls1_imgs.append(temp)#[print(i.shape) for i in ls1_imgs]ls1_shape = [l.shape for l in ls1_imgs]for i in range(n1,self.n_images):for l in range(0,layers):if i!=self.n_images-1:self.pyramids[i][l] = transform.resize(self.pyramids[i][l],(self.pyramids[i][l].shape[0],max(1,ls1_shape[l][1]//n2)))else:w = ls1_shape[l][1] - ls1_shape[l][1]//n2*(n2-1)#h = ls1_shape[l][0] - ls1_shape[l][0]//n2*(n2-1)self.pyramids[i][l] = transform.resize(self.pyramids[i][l],(self.pyramids[i][l].shape[0],w))[print(i[0].shape) for i in self.pyramids]ls2_imgs = []for l in range(0,layers):temp = self.pyramids[n1][l]for i in range(n1+1,self.n_images):temp = np.hstack((temp[:,:],self.pyramids[i][l][:,:]))ls2_imgs.append(temp)self.ls = [][print(i.shape) for i in ls1_imgs][print(i.shape) for i in ls2_imgs]for l1,l2 in zip(ls1_imgs,ls2_imgs):l = np.vstack((l1[:,:],l2[:,:]))self.ls.append(l)print(len(self.ls))[print(i.shape) for i in self.ls]def sfusion(self):new_image = self.ls[-1]for i in range(len(self.ls) - 1):temp_img = self.ls[len(self.ls) - 2 - i]# gs_tmp = GS[len(LS)-2-i]# z = temp_img.shape[0]# o = temp_img.shape[1]# new_image = transform.resize(new_image,(rows,cols))new_image = transform.rescale(new_image, 2, multichannel=True)temp_img = transform.resize(temp_img, (new_image.shape[0], new_image.shape[1]))# gs_tmp = transform.resize(gs_tmp, (new_image.shape[0], new_image.shape[1]))new_image = new_image + temp_imgself.fusion_image = new_imageplt.imshow(new_image*3)plt.show()def CollageCreate(AddressofFolder):f = Fusion(AddressofFolder)print([i.shape for i in f.images])f.sfusion()if __name__ == '__main__':CollageCreate('images')

代码结果示例

图像融合 skimage相关推荐

图像融合质量评价方法SSIM、PSNR、EN、MSE与NRMSE（一）
文章目录 1 前言 2 融合评价指标介绍 2.1 结构相似性 SSIM 2.2 峰值信噪比 PSNR 2.3 信息熵 Entropy 2.4 均方误差 MSE 2.5 归一化均方根误差 NRMSE 3 ...
图像融合质量评价方法MSSIM、MS-SSIM、FS、Qmi、Qabf与VIFF（三）
文章目录 1 前言 2 融合评价指标介绍 2.1 结构相似性(SSIM)相关 2.2 互信息(MI)相关 2.3 基于视觉信息保真度的指标VIFF 2.4 融合质量Qabf 3 代码实现 3.1 多层 ...
【opencv】(2) 图像处理：边界填充、图像融合、图像阈值、数值计算
主要内容有:边界填充 cv2.copyMakeBorder(),数值计算 cv2.add(),改变尺寸 cv2.resize(),图像融合 cv2.addWeighted(),图像阈值 cv2.thr ...
python——图像处理3（均值偏移、改变亮度、图像修复、图像融合）
https://blog.csdn.net/gm_ergou/article/details/92846396 1.均值偏移(磨皮效果) import cv2 as cv import numpy a ...
matlab将二值图像与原图重叠_[转载]图像处理matlab及图像融合图像镶嵌图像拼接...
在实际的对图像处理过程中,由于我们读出的图像是unit8型,而在MATLAB的矩阵运算中要求所有的运算变量为double型(双精度型).因此读出的图像数据不能直接进行相加求平均,因此必须使用一个函数将 ...
matlab图像融合评价,MATLAB 图像融合评估算法
MATLAB 图像融合评价算法 function laplacian(ori_A,ori_B) path(path,'FusionEvaluation/') %Ori_A = imread('cloc ...
图像拼接和图像融合技术
from: OpenCV探索之路(二十四)图像拼接和图像融合技术图像拼接在实际的应用场景很广,比如无人机航拍,遥感图像等等,图像拼接是进一步做图像理解基础步骤,拼接效果的好坏直接影响接下来的工作,所 ...
泊松图像融合(泊松融合)
泊松图像融合(泊松融合) from: http://blog.csdn.net/baimafujinji/article/details/46787837 在之前的文章中,我们详细介绍了基于泊松方程的 ...
基于权因子的图像融合
基于权因子的图像融合参见: OpenCV alpha(权因子) 融合举例

图像融合 skimage

图像融合skimage

图像融合

代码结果示例

图像融合 skimage相关推荐

最新文章

热门文章