43、一种图像边缘粗糙度计算方法

Bao@Ting

已于 2024-06-20 12:57:48 修改

阅读量2.7k

点赞数 2

分类专栏：软件文章标签： matlab 图像边缘粗糙度粗糙度

于 2021-12-19 22:13:37 首次发布

本文链接：https://blog.csdn.net/baotting/article/details/122030357

版权

软件专栏收录该内容

14 篇文章 1 订阅

订阅专栏

一、简介

二、问题引入

问题：使用自己的方法求如下四张图片的边缘粗糙度？

三、Matlab代码

原理分析：

        步骤一、导入原图
  步骤二、图像二值化
  步骤三、求最大连通图
  步骤四、进行巴特沃斯滤波
  步骤五、求原图像边缘长度
  步骤六、求滤掉的面积
步骤七、相除得到单位长度滤掉的面积

%用单位长度上被平滑掉的面积刻画闭合图形的边缘粗糙度(尽量让图形布满视野)
%对于非连通图像，让图像的两端与图像的左右边界重合
clear;clc;
%原图处理
ImageSource = imread('E:\software\MatLab2020b\myWorkPlace\linex.png');%原图
Image_Binary = im2bw(ImageSource,graythresh(ImageSource));%二值图
Image_Max = BigestConnectedDomain(Image_Binary);%最大连通图
[height,width] = size(Image_Binary);%图片的尺寸
[ImageSourceEdge,count_o,border_o] = ImageExcision(ImageSource,Image_Max,1);
%布特沃兹滤波图
n = 3;
d = 0.025*height;
im_out_b = ButterworthFilter(Image_Max,'l',n,d); 
im_out_bw = im2bw(im_out_b,graythresh(im_out_b));
[ImageButterEdge,count_n,border_n] = ImageExcision(ImageSource,im_out_bw,1);
%显示图形
[n2,m2] = size(border_n);
ImageSourceMax = zeros(height,width,3,'uint8');
ImageButterMax = zeros(height,width,3,'uint8');
ImageSourceMax(:,:,1) = Image_Max.*255;
ImageButterMax(:,:,2) = im_out_bw.*255;
ImageAdd = ImageButterMax+ImageSourceMax;
figure(1);
subplot(2,2,1);imshow(ImageSource);title('原图');
subplot(2,2,2);imshow(Image_Binary);title('二值图');
subplot(2,2,3);imshow(Image_Max);title('最大连通图');
subplot(2,2,4);imshow(ImageSourceEdge);title('原始边缘标定');
figure(2);
subplot(2,2,1);imshow(im_out_b);title('巴特沃兹滤波');
subplot(2,2,2);imshow(ImageSourceMax);title('巴特沃兹滤波二值图');
subplot(2,2,3);imshow(ImageButterMax);title('滤波边缘标定');
subplot(2,2,4);imshow(ImageSourceMax+ImageButterMax);title('滤波前后叠加');
%求粗糙度
sumx = sum(sum(ImageAdd(:,:,1)==255&ImageAdd(:,:,2)==0&ImageAdd(:,:,3)==0))+sum(sum(ImageAdd(:,:,1)==0&ImageAdd(:,:,2)==255&ImageAdd(:,:,3)==0));
ronghness = sumx/n2;
fprintf("总面积 = %d,单位长度面积 = %f\n",sumx,ronghness);

%函数1、求最大连通图
function [output_image,area] = BigestConnectedDomain(image)
%本函数求二值化的图像的最大连通域
input_image = imfill(image,'hole');%填充连通域内孤立的孔洞
%获取连通域信息
[L,num] = bwlabel(input_image,4);
x = zeros(1,num);
for i = 1:num
    x(i) = sum(sum(L == i));
end
%定位最大单连通域的位置
[~,n] = max(x);
%输出图像
output_image = (L == n);
%输出最大单连通域的面积
area = sum(sum(output_image));
end
%函数2、巴特沃斯滤波器
function [ output_image ] = ButterworthFilter( image ,mold,n,d)
%本函数为巴特沃斯滤波器，用于对图像进行巴特沃斯滤波
%   image参数为读入的图像矩阵；
%   mold参数为巴特沃斯滤波器的工作模式，有‘l’与‘h’可选，l为低通滤波，h为高通滤波；
%   n参数为巴特沃斯滤波器的阶数；
%   d参数为巴特沃斯滤波器的截止频率。
    hight=size(image,1);
    wide=size(image,2);

    fftim = fftshift(fft2(double(image)));
    [x,y]=meshgrid(floor(-wide/2):floor(wide/2)-1,floor(-hight/2):floor(hight/2)-1);

    B = sqrt(2) - 1; %定义巴特沃斯滤波器中的参数B
    D = sqrt(x.^2 + y.^2); %计算各点到图像中心的距离
    %定义滤波器函数
    if mold == 'l'
        hhp = 1 ./ (1 + B * ((D ./ d).^(2 * n)));
    elseif mold == 'h'
        hhp = 1 ./ (1 + B * ((d ./ D).^(2 * n)));
    end
    %将滤波器的矩阵维度与图像的矩阵维度匹配
    [~,~,page]= size(image);
    hhp1 = hhp;
    i = 1;
    while i < page
        hhp1 = cat(3,hhp1,hhp);
        i = i + 1;
    end 
    hhp = hhp1;
    %滤波
    out_spec_centre = fftim .* hhp;
    %象限转换
    out_spec = ifftshift(out_spec_centre);
    %反变换并取实部
    out = real(ifft2(out_spec));
    %归一化
    out = (out - min(out(:))) / (max(out(:)) - min(out(:)));
    %反归一化图像
    output_image = uint8(255*out);
end
%函数3、output_image:原图标记边缘，border_length:边缘点数，border_location:边缘点坐标
function [output_image,border_length,border_location] = ImageExcision(original_image,bw_image,color_type)
im = original_image;
[row,col] = size(bw_image);%row:高度，col宽度
%-------------------------------------------------------------------------%
count = 0;  %边界点的个数
b_loc = []; %所有边界点的坐标
for j = 2:row-1 %内部像素边界判断
    for k = 2:col-1
        if(bw_image(j,k)==1)
           if (bw_image(j-1,k)== 0 || bw_image(j+1,k)== 0 || bw_image(j,k-1)== 0 ||bw_image(j,k+1)== 0)
               if(color_type == 1) 
                   im(j,k,1) = 255;im(j,k,2) = 0;im(j,k,3) = 0;
               else
                  im(j,k,1) = 0;im(j,k,2) = 0;im(j,k,3) = 255;
               end
               b_loc = [b_loc;[j k]];
               count = count +1; 
           end
        end
    end
end
%-------------------------------------------------------------------------%
%图像边缘像素边界判断(四个角上的像素点除外)
%第一行与最后一行像素
for i = 2:col-1
    if(bw_image(1,i) == 1)
        if(bw_image(1,i+1) == 0 || bw_image(1,i-1) == 0 || bw_image(2,i) == 0)
            if(color_type == 1) 
                im(1,i,1) = 255;im(1,i,2) = 0;im(1,i,3) = 0;
            else
                im(1,i,1) = 0;im(1,i,2) = 0;im(1,i,3) = 255;
            end
            b_loc = [b_loc;[1 i]];
            count = count +1;
        end
    end
    if(bw_image(row,i) == 1)
        if(bw_image(row,i+1) == 0 || bw_image(row,i-1) == 0 || bw_image(row-1,i) == 0)
            if(color_type == 1) 
                im(row,i,1) = 255;im(row,i,2) = 0;im(row,i,3) = 0;
            else
                im(row,i,1) = 0;im(row,i,2) = 0;im(row,i,3) = 255;
            end
            b_loc = [b_loc;[row i]];
            count = count +1;
        end
    end
end
%第一列与最后一列像素
for i = 2:row-1
    if(bw_image(i,1) == 1)
        if (bw_image(i+1,1) == 0 || bw_image(i-1,1) == 0 || bw_image(i,2) == 0)
            if(color_type == 1) 
                im(i,1,1) = 255;im(i,1,2) = 0;im(i,1,3) = 0;
            else
                im(i,1,1) = 0;im(i,1,2) = 0;im(i,1,3) = 255;
            end
            b_loc = [b_loc;[i 1]];
            count = count +1;
        end
    end
    if(bw_image(i,col) == 1)
        if (bw_image(i+1,col) == 0 || bw_image(i-1,col) == 0 || bw_image(i,col-1) == 0)
            if(color_type == 1) 
                im(i,col,1) = 255;im(i,col,2) = 0;im(i,col,3) = 0;
            else
                im(i,col,1) = 0;im(i,col,2) = 0;im(i,col,3) = 255;
            end
            b_loc = [b_loc;[i col]];
            count = count +1;
        end
    end
end
%-------------------------------------------------------------------------%
%四个角上的像素点判断
if(bw_image(1,1)==1)%左上点
    if (bw_image(1,2) == 0 || bw_image(2,1) == 0)
        im(i,col,1) = 255;im(i,col,2) = 0;im(i,col,3) = 0;
        if(color_type == 1) 
            im(i,col,1) = 255;im(i,col,2) = 0;im(i,col,3) = 0;
        else
            im(i,col,1) = 0;im(i,col,2) = 0;im(i,col,3) = 255;
        end       
        b_loc = [b_loc;[1 1]];
        count = count +1;
    end
end         
if(bw_image(row,1)==1)%左下点
    if(bw_image(row,2) == 0 || bw_image(row-1,1) == 0)        
        if(color_type == 1) 
            im(i,col,1) = 255;im(i,col,2) = 0;im(i,col,3) = 0;
        else
            im(i,col,1) = 0;im(i,col,2) = 0;im(i,col,3) = 255;
        end  
        b_loc = [b_loc;[row 1]];
        count = count +1;
    end
end
        
if(bw_image(1,col) == 1)%右上点
    if(bw_image(1,col-1) == 0 || bw_image(2,col) == 0)
        if(color_type == 1) 
            im(i,col,1) = 255;im(i,col,2) = 0;im(i,col,3) = 0;
        else
            im(i,col,1) = 0;im(i,col,2) = 0;im(i,col,3) = 255;
        end  
        b_loc = [b_loc;[1 col]];
        count = count +1;
    end
end     
if(bw_image(row,col)==1) %右下点
    if(bw_image(row,col-1) == 0 || bw_image(row-1,col) == 0)
        if(color_type == 1) 
            im(i,col,1) = 255;im(i,col,2) = 0;im(i,col,3) = 0;
        else
            im(i,col,1) = 0;im(i,col,2) = 0;im(i,col,3) = 255;
        end  
        b_loc = [b_loc;[row col]];
        count = count +1;
    end
end
output_image = im;
border_length = count/2;
border_location = b_loc; %所有边界点坐标
end
function ret=fun_edgex(Input_image) %zhoujinhao add (提取图像的边缘)
    ret_img = Input_image;
    [height,width]=size(Input_image);
    for i_x = 1:(height)
        for j_x = 1:(width) %扫描每一个像素
            if (i_x>1&&i_x<height&&j_x>1&&j_x<width)
                if Input_image(i_x,j_x)==1
                    ret_img(i_x,j_x)=~(Input_image(i_x-1,j_x)&&Input_image(i_x+1,j_x)&&Input_image(i_x,j_x-1)&&Input_image(i_x,j_x+1));
                end
            end
        end
    end
    ret = ret_img;
end