《玩儿起来吧》MATLAB 实时图像处理（五）引体计数器

用MATLAB实现的运动计数器。大家学了图像处理之后，是不是就想学以致用。而要用于生活中，最好的方式就是和摄像头搭配，实现实时图像处理。今天就给大家分享MATLAB实时图像处理的基础方法，希望能帮你快速入门。

by 今天不飞了

---

视频教学地址

《玩儿起来吧》MATLAB实时图像处理（五）引体计数器

---

代码

核心函数findpeaks()

function SportCount()

%% 创建GUI
Fig = figure('Position',[500,450,980,500]);

Pnl1 = uipanel(Fig,'Position',[0.05,0.17,0.9,0.8]);
Pnl2 = uipanel(Fig,'Position',[0.05,0.05,0.9,0.1]);

Axes1 = axes(Pnl1,'Position',[0,0,1/2,1]);
Axes2 = axes(Pnl1,'Position',[1/2,0,1/2,1]);

Bt = uicontrol(Pnl2,'style','togglebutton','String','锁定目标并开始计数','Fontsize',16,...
    'Units','normalized','Position',[0,0,0.5,1],'Callback',@LockTarget);
Txt = uicontrol(Pnl2,'style','text','String',[],'Fontsize',16,...
    'Units','normalized','Position',[0.5,0,0.5,1]);
drawnow

%% 开启摄像头
Hcamera = [];
Hobj = [];
if isempty(Hcamera)
    Hobj = videoinput('winvideo',1,'MJPG_640x480');
    Hcamera = preview(Hobj);
    frame = getsnapshot(Hobj);  % 获取帧
end

%% 初始化参数
flag = 0;
tracker = vision.PointTracker('MaxBidirectionalError',1);
gap = 1;
objectRegion = [0,0,0,0];
objFrame = [];
points = [];

% 初始化计数
rows = 0;
bodyPosition = [];
time = 0;
H = fspecial('average',[1,5]);
MPP = 100;
count = 0;

%% 开始
while 1
    if ishandle(Hcamera)
        % 获取影像
        frame = getsnapshot(Hobj);  % 获取帧
        if flag
            % 检测
            body(frame);
            imshow(frame,'Parent',Axes1)
        else
            % 普通显示
            imshow(frame,'Parent',Axes1)
        end
        drawnow
    else
        break
    end
end

    function LockTarget(~,~)
        bodyPosition = [];
        time = 0;
        count = 0;
        set(Txt,'String',num2str(count))
        % 降采样
        frame = frame(1:gap:end,1:gap:end,:);
        objFrame = frame;
        % 确定目标位置
        [x,y,~] = ginput(2);
        objectRegion = [min(x),min(y),max(x)-min(x),max(y)-min(y)]/gap;
        % 检测特征点
        points = detectMinEigenFeatures(rgb2gray(frame),'ROI',objectRegion);        
        % 初始化跟踪器
        release(tracker)
        initialize(tracker, points.Location, frame);
        flag = 1;
        rows = size(frame,1);
        bodyPosition = ones(1,30*3600)*(rows-mean(points.Location(:,2),1));        
    end

    function body(frame)
        % 降采样
        frame = frame(1:gap:end,1:gap:end,:);
        % 追踪
        [point,validity] = tracker(frame);
        point = point(validity,:);
        % 更新目标
        if size(point,1)<10
            release(tracker)
            initialize(tracker, points.Location, objFrame);
            [point,validity] = tracker(frame);
            point = point(validity,:);
        end
        % 反算坐标
        point = point*gap;
        
        time = time+1;
        bodyPosition(time) = rows-mean(point(:,2),1);
        tmp = imfilter(bodyPosition(1:time+10),H,'replicate','same');
        [pks, locs] = findpeaks(tmp, 'MinPeakProminence', MPP); 
        count = length(pks);
        
        % 显示
        plot(Axes2,tmp,'b-')
        hold(Axes2,'on')
        plot(Axes2,locs,pks,'r<')
        hold(Axes2,'off')
        axis(Axes2,[0,time+30,0,rows])
        set(Txt,'String',num2str(count))
    end

end

---

其他

1. 欢迎大家自行修改为“自动定位人物并开始计数”。
2. MATLAB建议2018及以上版本。