[VP] 拉普拉斯算子 Laplacian Filter

---

拉普拉斯算子

长这样：

$$\left[\begin{array}{ccc}-1& 2& -1\\ -1& 2& -1\\ -1& 2& -1\end{array}\right]$$

性质

Image = np.array([[0, 0, 0, 0, 0, 0], 
                  [0, 1, 5, 9, 7, 0], 
                  [0, 1, 5, 9, 7, 0], 
                  [0, 1, 5, 9, 7, 0], 
                  [0, 1, 5, 9, 7, 0], 
                  [0, 0, 0, 0, 0, 0]])

Kernel = [[-1, 2, -1], [-1, 2, -1], [-1, 2, -1]]

ax1 = plt.subplot(1,2,1)
plt.imshow(Image)
plt.title('Image')

ax1 = plt.subplot(1,2,2)
plt.imshow(Kernel)
plt.title('Kernel')


def convolution(Image, Kernel):
    count = 0
    for i in range(0, 3):
        for j in range(0, 3):
            count += Image[2-i][2-j]*Kernel[i][j]
    return count

def correlation(Image, Kernel):
    count = 0
    for i in range(0, 3):
        for j in range(0, 3):
            count += Image[i][j]*Kernel[i][j]
    return count


result1 = [[0, 0, 0, 0], 
          [0, 0, 0, 0],
          [0, 0, 0, 0], 
          [0, 0, 0, 0]]

for i in range(4):
    for j in range(4):
        result1[i][j] = convolution(Image[i:i+3, j:j+3], Kernel)


result2 = [[0, 0, 0, 0], 
          [0, 0, 0, 0],
          [0, 0, 0, 0], 
          [0, 0, 0, 0]]

for i in range(4):
    for j in range(4):
        result2[i][j] = correlation(Image[i:i+3, j:j+3], Kernel)


_, ax = plt.subplots(1, 2, figsize = (10, 10))

ax[0].imshow(result1)
ax[0].set_title('Convolution Result')
ax[1].imshow(result2)
ax[1].set_title('Correlation Result')

从结果看出，用拉普拉斯算子对图像做卷积和关联操作，其结果是一样的，因为拉普拉斯算子本身旋转 180° 的结果还是它本身。