python小记

发现：

目前来说，opencv，numpy，PIL之间的图像数据可以不经过转换直接使用（需要注意的是，opencv的图像一般都是BGR，其他都是RGB）

• 计算平均值

def get_average(records):
    """
    平均值
    """
    return sum(records) / len(records)

• 计算方差

def get_variance(records):
    """
    方差 反映一个数据集的离散程度
    """
    average = get_average(records)
    return sum([(x - average) ** 2 for x in records]) / len(records)

• opencv上画圆点

def puttext(x, y, img):
    text = '(' + str(x) + ',' + str(y) + ')' 
    org = (x, y) # 文字左下角坐标
    fontFace = cv2.FONT_HERSHEY_COMPLEX
    fontScale = 0.3
    fontcolor = (0, 255, 255)  # BGR
    thickness = 1
    lineType = 8
    cv2.putText(img, text, org, fontFace, fontScale, fontcolor, thickness, lineType)
    return 0

• opencv边缘提取

def find_draw_contours(final_image_cv, num):
	# final_image_cv 是mat格式的opencv图像
	# num 是决定画的边缘线的颜色，1是白，否则是黑色
    if final_image_cv.shape[2] == 3:
        gray = cv2.cvtColor(final_image_cv, cv2.COLOR_BGR2GRAY)
        ret, binary = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY)
    else:
        ret, binary = cv2.threshold(final_image_cv, 0, 255, cv2.THRESH_BINARY)
    contours, hierarchy = cv2.findContours(binary, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    if num:
        color_ = [255, 255, 255]
    else:
        color_ = [0, 0, 0]
    cv2.drawContours(final_image_cv, contours, -1, color_, 3)
    return final_image_cv

• numpy数据类型转换

    Y = Y.astype(np.float64)

• opencv图像重叠(经过检测，两个思路，思路二效率是前者的6倍)：

思路1，使用PIL的Image.blend:

def blend_1(img1, img2):
    '''
    Function：
        利用PIL，将img1和img2进行重叠合并
    Para:
        img1是原始图像
        img2是预测后经过处理的最终预测图像
    '''
    s_time = time.time()

    im1 = Image.fromarray(cv2.cvtColor(img1, cv2.COLOR_BGR2RGB)) # opencv转PIL
    im2 = Image.fromarray(cv2.cvtColor(img2, cv2.COLOR_BGR2RGB)) # opencv转PIL
    im = Image.blend(im1, im2, 0.5)                              # 重叠
    img = cv2.cvtColor(numpy.asarray(im), cv2.COLOR_RGB2BGR)     # PIL转opencv 
    u_time = time.time()
    img_time = u_time - s_time
    print("time: {} ".format(img_time))
    return img # 返回合并opencv图像

思路2，使用cv2.addWeight（good）:

def blend_2(img1, img2):
    '''
    Function：
        利用PIL，将img1和img2进行重叠合并
    Para:
        img1是原始图像
        img2是预测后经过处理的最终预测图像
    '''
    s_time = time.time()
    dst = cv2.addWeighted(img1,0.5,img2,0.5,0)
    u_time = time.time()
    img_time = u_time - s_time
    print("time: {} ".format(img_time))
    return dst # 返回合并opencv图