程序

模拟光斑

# 理想光斑模型
def Gaussian2D(sigma, miu_1, miu_2, x, y):
    z = (1 /sigma) * np.exp(-((x - miu_1) ** 2 + (y - miu_2) ** 2) / (sigma ** 2))
    return z
# 生成带有高斯噪声光斑模型
#创建光斑模型图片
def generate(img_shape, sigma, miu_1, miu_2,N):
    (width, height) = img_shape
    X = np.arange(0, width, 1)
    Y = np.arange(0, height, 1)
    X, Y = np.meshgrid(X, Y)
    img = Gaussian2D(sigma, miu_1, miu_2, X, Y)
    # img = ns.salt_noise(img,N)
    # img = img+ns.gau_noise(img,0.005)
    img = ns.salt_noise(img, N)
    fig = plt.figure()
    ax = fig.gca(projection='3d')
    surf = ax.plot_surface(X, Y, img, cmap=cm.coolwarm,
                           linewidth=0, antialiased=False)
    ax.zaxis.set_major_locator(LinearLocator(10))
    ax.zaxis.set_major_formatter(FormatStrFormatter('%.02f'))
    # Add a color bar which maps values to colors.
    fig.colorbar(surf, shrink=0.5, aspect=5)
    plt.show()
    return img

模拟随机的椒盐噪声

def salt_noise(img,N):
    x_sum = 0
    y_sum = 0
    for i in range(N):
        x = np.random.randint(0,500)
        y = np.random.randint(0,500)
        img[y][x] = np.max(img)
    return img

通过CMOS相机得到的灰白光斑照片

def img_read(img_path):
    img = plt.imread(img_path)
    #+++++++++++++++++++++归一化到0-255之间+++++++++++++++
    if len(img.shape) > 2:
        img = img.mean(axis=2)
    originalMinValue = np.min(img)
    originalMaxValue = np.max(img)
    originalRange = originalMaxValue - originalMinValue
    img = 255 * (img - originalMinValue) / originalRange
    img = np.around(img)
    print(img)
    return img

质心算法

def tr_centroid_algorithm(shape,img):
    # 灰度值加总
    x, y = shape
    threshold = np.sum(img)/(x*y)
    grayX_valueSum = 0.0
    grayY_valueSum = 0.0
    gray_valueSum = 0.0
    for i in range(x):
        for j in range(y):
            if img[i][j] < threshold:
                img[i][j] = 0
            grayX_valueSum += i * img[i][j]
            grayY_valueSum += j * img[i][j]
            gray_valueSum += img[i][j]
    x_c = grayX_valueSum / gray_valueSum
    y_c = grayY_valueSum / gray_valueSum
    center_position = (y_c, x_c)
    return center_position
#=================加权平方质心算法================
##不适合椒盐噪声比较大的
def square_centroid_algorithm(shape, img):
    #加权平方质心算法
    x, y = shape
    grayX_valueSum = 0.0
    grayY_valueSum = 0.0
    gray_valueSum = 0.0
    threshold = np.sum(img) / (x * y)
    for i in range(x):
        for j in range(y):
            if img[i][j] < threshold:
                img[i][j] = 0
            grayX_valueSum += i*((img[i][j])**2)
            grayY_valueSum += j*((img[i][j])**2)
            gray_valueSum += (img[i][j])**2
    x_c = grayX_valueSum/gray_valueSum
    y_c = grayY_valueSum/gray_valueSum
    center_position = (y_c, x_c)
    return center_position

主程序

if __name__ == '__main__':
    path = "008.png"
    img = er.img_read(path)
    shape = img.shape
    center_ax1 = ca.tr_centroid_algorithm(shape,img)
    center_ax2 = ca.square_centroid_algorithm(shape, img)
    print(center_ax1,center_ax2)
    plt.figure()
    plt.plot(center_ax1[0],center_ax1[1],marker='o',markersize=5,
             label="traditional centroid algorithm")
    plt.plot(center_ax2[0],center_ax2[1],marker='o',markersize=2,
             label="square centroid algorithm")
    plt.imshow(img,cmap="gray")
    plt.legend()
    plt.show()

仿真结果

理想高斯光斑对应的实验结果(289，307)

加上随机椒盐噪声(250,250)

传统(249.737741615272, 250.04379953388042)
加权(249.49177547117412, 250.08141883380614)

传统(249.9948760536369, 249.95908029324465)
加权(249.99246537464057, 249.9253089391913)

实验结果

实验原理图

实验装置图

实验结果处理

CMOS像素：2048×1536

001.pgm

传统阈值算法：(946.8248700956387, 739.0951441664839)
平方加权算法：(946.9951532351402, 739.6160210844841)

002.pgm

传统阈值算法：(1219.9349157580625, 432.73786390104505)
平方加权算法：(1222.6862424084402, 428.7885181805899)

003.pgm

传统阈值算法：(1158.2684591625016, 996.4485059513937)
平方加权算法：(1159.0742006690245, 1004.2174495319779)

004.pgm

传统阈值算法：(1615.910249562971, 511.5607608509442)
平方加权算法：(1628.8798166031754, 491.8004277438292)

备注：001.pgm和002.pgm是激光器直接照射，003.pgm和004.pgm是上述装置图得到的结果