MATLAB中使用HSV阈值分割交通标志

前言

通常，我们会根据颜色的不同分割不同的目标，例如我们日常生活中经常看到的交通标志，一个很直观的想法就是将颜色从RGB空间转换到HSV空间，根据色调（H）或者饱和度分割不同的目标，下面就是这样一个例子；

代码

%%
clc;clear;close all
%%
img = imread('test3.jpg');
imgHSV = rgb2hsv(img);
imgt = imgHSV(:,:,2);
%首先进行饱和度筛选（0到1之间），越大说明选择条件越严格
BW = imgt>0.8;
CC = bwconncomp(BW);
boxs = regionprops(CC,'BoundingBox');
flag = false(size(boxs));
for i = 1:CC.NumObjects
    if numel(CC.PixelIdxList{i})>200%调整这个值可自定义目标区域的尺寸
        flag(i) = 1;
    end
end
boxs(~flag) = [];
% boxs = floor(boxs.BoundingBox);
%根据图像里的H分量进行判断
ind = find(flag);
for i = 1:numel(boxs)
    H = imgHSV(:,:,1);
    pixels = mean(H(CC.PixelIdxList{ind(i)}));
    if pixels< 0.2
        res1 = false([size(img,1),size(img,2)]);
        res1(CC.PixelIdxList{ind(i)}) = 1;
        imwrite(res1,['imgY_',num2str(i),'.png']);
        bd = floor(boxs(i).BoundingBox);
        maskedImg  = img(bd(2):bd(2)+bd(4),bd(1):bd(1)+bd(3),:);
        imwrite( maskedImg,['imgYMasked_',num2str(i),'.png']);
    end
    if pixels> 0.5 && pixels<0.7
        res1 = false([size(img,1),size(img,2)]);
        res1(CC.PixelIdxList{ind(i)}) = 1;
        imwrite(res1,['imgB_',num2str(i),'.png']);
        bd = floor(boxs(i).BoundingBox);
        maskedImg  = img(bd(2):bd(2)+bd(4),bd(1):bd(1)+bd(3),:);
        imwrite( maskedImg,['imgBMasked_',num2str(i),'.png']);
    end
    if pixels>0.8
        res1 = false([size(img,1),size(img,2)]);
        res1(CC.PixelIdxList{ind(i)}) = 1;
        imwrite(res1,['imgR_',num2str(i),'.png']);
        bd = floor(boxs(i).BoundingBox);
        maskedImg  = img(bd(2):bd(2)+bd(4),bd(1):bd(1)+bd(3),:);
        imwrite( maskedImg,['imgRMasked_',num2str(i),'.png']);
    end
end
% BW = imbinarize(imgt);
% imshow(BW)
% imshow(imgHSV(:,:,2),[])

示例图像