python np.fliplr三通道图像与图像投影

在彩色图像上画一条直线：

   for item in lc_imglists:
       img_ori = cv2.imread(item)

       cv2.line(img_ori,(100,100),(1000,1000),(0,0,255),3)

       imshow(img_ori,600,600,'img',True)

如果对原图像上下，或左右翻转后再画直线，第一想法是：

    for item in lc_imglists:
        img = cv2.imread(item)[:,0:2000,:]
        img = np.flipud(img)  # np.fliplr
        cv2.line(img,(100,100),(1000,1000),(0,0,255),3)
        imshow(img,600,600,'img',True)

图像中没有显示所画的红线，退而求其次，对每个通道进行翻转，再进行拼接：

    for item in lc_imglists:
        img = cv2.imread(item)[:,0:2000,:]

        flag = True
        if flag:
            b,g,r = cv2.split(img)
            b = np.fliplr(b)
            g = np.fliplr(g)
            r = np.fliplr(r)
            img[:,:,0] = b
            img[:,:,1] = g
            img[:,:,2] = r
        cv2.line(img,(100,100),(1000,1000),(0,0,255),3)

        imshow(img,600,600,'img',True)

水平 竖直投影法在图像分割中应用广泛, 代码实现如下:

def CalcImgProjection(img):
    rowProjection = np.mean(img, axis=0)
    colProjection = np.mean(img, axis=1)

    plt.figure()

    plt.subplot(121)
    plt.imshow(img, 'gray')
    plt.title('img')

    plt.subplot(222)
    plt.plot(rowProjection, 'r*')
    plt.title('row')

    plt.subplot(224)
    plt.plot(colProjection, 'go')
    plt.title('col')

    plt.show()

if __name__ == '__main__':
    data = './datasets/lowcontrast_thread/'
    type = '*.bmp'
    imglists = glob.glob(data + type)

    for item in imglists:
        img = cv2.imread(item,0)[:,:2500]
        CalcImgProjection(img)

利用像素值的变化关系,选择合理的阈值,即可实现目标区域的分割.