Commit c8f88061 authored by Sparkf's avatar Sparkf 🏙️
Browse files

lab5 submit

parent 11c932ff

lab5/butterworse_notch.py

+28 −13
Changes for lab5/butterworse_notch.py: 28 added lines, 13 removed lines.
Original line number Diff line number Diff line 
import matplotlib.pyplot as plt

import numpy as np

from PIL import Image

import time

import lab5_lib





def butterworse_pass_11812418(input_img, sigma):

def butterworse_pass_11812418(input_img, sigma, n):

    # Reading of the image into numpy array:

    img = Image.open(input_img)

    # FFT transform for img arr
@@ -19,15 +18,25 @@ def butterworse_pass_11812418(input_img, sigma): 
    # 492,336



    # define butterworth

    butterworth_filter = 1-lab5_lib.generate_butterworth(40, 40, 2, sigma)

    bf_x = [70,150,310,395]

    bf_y = [90,210]

    butterworth_filter = np.ones([row, col])

    # = 1-lab5_lib.generate_butterworth(40, 40, 2, sigma)

    centers = [

        [109, 87],

        [109, 170],

        [115, 330],

        [115, 412],

        [227, 405],

        [227, 325],

        [223, 162],

        [223, 79]

    ]

    for point in centers:

        butterworth_filter = butterworth_filter - lab5_lib.generate_butterworth(row, col, n, sigma, point[0], point[1])



    for x in bf_x:

        for y in bf_y:

            input_image[x:x+40,y:y+40] = np.multiply(input_image[x:x+40,y:y+40],butterworth_filter)

    plt.imshow(np.log(np.abs(input_image)))

    plt.show()

    input_image = np.multiply(input_image, butterworth_filter)



    # plt.imshow(np.log(np.abs(input_image)))

    # plt.show()



    # plt.imshow(input_image)

    # plt.show()
@@ -39,12 +48,18 @@ def butterworse_pass_11812418(input_img, sigma): 
    # plt.imshow(np.real(output_img))

    # plt.show()



    return np.abs(output_img)

    return np.abs(output_img), np.log(np.abs(input_image))





if __name__ == '__main__':

    start_time = time.time()

    A = butterworse_pass_11812418("Q5_3.tif", 100)

    for sigma in [10, 40, 80, 120]:

        for n in [1, 2, 3, 4]:

            A, A_S = butterworse_pass_11812418("Q5_3.tif", sigma, n)

            op_image = Image.fromarray(A.astype(np.uint8))

            op_image.save("output/Q5_3_M_S" + str(sigma) + "_N" + str(n) + ".png")

            # plt.imshow(A_S)

            # plt.axis('off')

            # plt.savefig("output/Q5_3_M_F" + str(sigma) + "_N" + str(n) + ".png", bbox_inches='tight', pad_inches=0,

            #             dpi=150)

    print("--- %s seconds ---" % (time.time() - start_time))
 
    op_image.save("output/Q5_3_M.tif")

lab5/gaussian.py

+31 −26
Changes for lab5/gaussian.py: 31 added lines, 26 removed lines.
Original line number Diff line number Diff line 
import matplotlib.pyplot as plt

import time

import numpy as np

from PIL import Image

from joblib import Parallel, delayed

import time

import lab5_lib
@@ -19,39 +17,46 @@ def gaussian_pass_11812418(input_img, sigma): 
    input_image = np.fft.fft2(input_image)

    input_image = np.fft.fftshift(input_image)

    # define gaussian

    gaussian_filter = lab5_lib.generate_gaussian(row, col, sigma)

    gaussian_filter_HP = lab5_lib.generate_gaussian(row, col, sigma)

    gaussian_filter_LP = np.ones([row, col]) - lab5_lib.generate_gaussian(row, col, sigma)

    # print(gaussian_filter)

    # plt.imshow(gaussian_filter)

    # plt.show()

    filtered_img = np.multiply(input_image, gaussian_filter)

    output_img = np.fft.ifft2(np.fft.ifftshift(filtered_img))

    output_img = lab5_lib.transform_centering(output_img)

    # print(np.min(np.abs(output_img)))

    # output_img = lab5_lib.normalize(output_img)

    # plt.imshow(np.real(output_img))

    filtered_img_HP = np.multiply(input_image, gaussian_filter_HP)

    output_img_HP = np.fft.ifft2(np.fft.ifftshift(filtered_img_HP))

    output_img_HP = lab5_lib.transform_centering(output_img_HP)

    filtered_img_LP = np.multiply(input_image, gaussian_filter_LP)

    output_img_LP = np.fft.ifft2(np.fft.ifftshift(filtered_img_LP))

    output_img_LP = lab5_lib.transform_centering(output_img_LP)

    # print(np.min(np.abs(output_img_HP)))

    # output_img_HP = lab5_lib.normalize(output_img_HP)

    # plt.imshow(np.real(output_img_HP))

    # plt.show()



    return np.abs(output_img), np.abs(input_image_origin-output_img)

    return np.abs(output_img_HP), np.abs(output_img_LP), lab5_lib.normalize(

        gaussian_filter_LP) * 255, lab5_lib.normalize(gaussian_filter_HP) * 255





if __name__ == '__main__':

    start_time = time.time()

    A_LP,A_HP = gaussian_pass_11812418("Q5_2.tif", 30)

    for i in [30, 60, 160]:

        A_LP, A_HP, G_HP, G_LP = gaussian_pass_11812418("Q5_2.tif", i)

        op_image_LP = Image.fromarray(A_LP.astype(np.uint8))

    op_image_LP.save("output/Q5_2_LP30.tif")

        op_image_LP.save("output/Q5_2_LP" + str(i) + ".tif")

        op_image_HP = Image.fromarray(A_HP.astype(np.uint8))

    op_image_HP.save("output/Q5_2_HP30.tif")

    A_LP,A_HP = gaussian_pass_11812418("Q5_2.tif", 60)

    op_image_LP = Image.fromarray(A_LP.astype(np.uint8))

    op_image_LP.save("output/Q5_2_LP60.tif")

    op_image_HP = Image.fromarray(A_HP.astype(np.uint8))

    op_image_HP.save("output/Q5_2_HP60.tif")

    A_LP,A_HP = gaussian_pass_11812418("Q5_2.tif", 160)

        op_image_HP.save("output/Q5_2_HP" + str(i) + ".tif")

        op_image_GL = Image.fromarray(G_LP.astype(np.uint8))

        op_image_GL.save("output/Q5_2_LP" + str(i) + "_F.tif")

        op_image_GH = Image.fromarray(G_HP.astype(np.uint8))

        op_image_GH.save("output/Q5_2_HP" + str(i) + "_F.tif")

    print("--- %s seconds ---" % (time.time() - start_time))

    for i in [30, 60, 160]:

        A_LP, A_HP, G_HP, G_LP = gaussian_pass_11812418("Q5_2.tif", i)

        op_image_LP = Image.fromarray(A_LP.astype(np.uint8))

    op_image_LP.save("output/Q5_2_LP160.tif")

        op_image_LP.save("output/Q5_2_LP" + str(i) + ".png")

        op_image_HP = Image.fromarray(A_HP.astype(np.uint8))

    op_image_HP.save("output/Q5_2_HP160.tif")





    print("--- %s seconds ---" % (time.time() - start_time))

        op_image_HP.save("output/Q5_2_HP" + str(i) + ".png")

        op_image_GL = Image.fromarray(G_LP.astype(np.uint8))

        op_image_GL.save("output/Q5_2_LP" + str(i) + "_F.png")

        op_image_GH = Image.fromarray(G_HP.astype(np.uint8))

        op_image_GH.save("output/Q5_2_HP" + str(i) + "_F.png")

lab5/lab5_lib.py

+4 −3
Changes for lab5/lab5_lib.py: 4 added lines, 3 removed lines.
Original line number Diff line number Diff line
@@ -34,6 +34,7 @@ def transform_centering(input_img): 
def DFT_kernel(input_image_padded, kernel):

    sz = (input_image_padded.shape[0] - kernel.shape[0], input_image_padded.shape[1] - kernel.shape[1])

    DFT_kernel_1 = np.pad(kernel, (((sz[0] + 1) // 2, sz[0] // 2), ((sz[1] + 1) // 2, sz[1] // 2)), 'constant')

    DFT_kernel_1 = multiply_center(DFT_kernel_1)

    DFT_kernel_1_fft = np.fft.fft2(DFT_kernel_1)

    return DFT_kernel_1_fft
@@ -49,10 +50,10 @@ def generate_gaussian(a, b, sigma): 






def generate_butterworth(row, col, n, sigma):

def generate_butterworth(row, col, n, sigma, cr, cc):

    x, y = np.meshgrid(np.linspace(0, col - 1, col), np.linspace(0, row - 1, row))

    x = x - col / 2

    y = y - row / 2

    x = x - cr

    y = y - cc

    d = np.sqrt(x * x + y * y)

    h = 1 / ((1 + (d / sigma)) ** (2 * n))

    return h

lab5/sobel.py

+19 −9
Changes for lab5/sobel.py: 19 added lines, 9 removed lines.
Original line number Diff line number Diff line
@@ -47,15 +47,26 @@ def sobel_11812418(input_img): 
    img_arr = lab5_lib.multiply_center(img_arr)



    img_arr_fft = np.fft.fft2(img_arr)

    # plt.imshow(20*np.log(np.abs((np.fft.fft2(img_arr)))))

    # plt.show()



    plt.imshow(20 * np.log(np.abs((np.fft.fft2(img_arr)))))

    heatmap = plt.pcolor(20 * np.log(np.abs((np.fft.fft2(img_arr)))))

    plt.colorbar(heatmap)

    plt.savefig('output/Q5_1_F.png', dpi=600)



    kernel1 = np.array([[-1, 0, 1], [-2, 0, 2], [-1, 0, 1]])

    kernel2 = np.array([[-1, -2, -1], [0, 0, 0], [1, 2, 1]])

    kernel1 = kernel1 + kernel2



    DFT_kernel_1_fft = lab5_lib.DFT_kernel(img_arr, kernel1)

    DFT_kernel_fft = DFT_kernel_1_fft



    # plt.imshow(np.abs(DFT_kernel_1_fft))

    # heatmap = plt.pcolor(np.abs(DFT_kernel_1_fft))

    # plt.colorbar(heatmap)

    # plt.savefig('output/sobel_kernel.png', dpi=600)



    # K1

    filtered_1 = np.real(np.fft.ifft2(img_arr_fft * DFT_kernel_1_fft))

    filtered_1 = np.real(np.fft.ifft2(img_arr_fft * DFT_kernel_fft))

    filtered_1 = lab5_lib.FFT_extract_padding(filtered_1)

    filtered_1 = lab5_lib.multiply_center(filtered_1)

    # plt.imshow(filtered_1)
@@ -65,14 +76,11 @@ def sobel_11812418(input_img): 
    # op_img = np.fft.ifft2(conv(img_arr, kernel_flat))

    # print(filtered_1)

    op_img = lab5_lib.normalize(filtered_1) * 255

    op_img = np.clip(op_img, 115, 145)

    op_img = lab5_lib.normalize(op_img)*255

    print(op_img)



    plt.imshow(op_img)

    plt.show()



    op_img_spacial = conv(input_image_origin,kernel1.flatten())

    op_img_spacial = conv(input_image_origin, np.flip(kernel1.flatten()))

    op_img_spacial = lab5_lib.normalize(op_img_spacial) * 255



    return op_img, op_img_spacial
@@ -83,5 +91,7 @@ A, A_S = sobel_11812418("Q5_1.tif") 
op_image = Image.fromarray(A.astype(np.uint8))

op_image_S = Image.fromarray(A_S.astype(np.uint8))

print("--- %s seconds ---" % (time.time() - start_time))

op_image.save("output/Q5_1_M.tif")

op_image_S.save("output/Q5_1_M_S.tif")
 
op_image.save("output/Q5_1_M_Frequency.tif")

op_image_S.save("output/Q5_1_M_Spatial.tif")

op_image.save("output/Q5_1_M_Frequency.png")

op_image_S.save("output/Q5_1_M_Spatial.png")

lab5/submit/EE326_Lab_Report_4.pdf

0 → 100644
+8.2 MiB

File added.

No diff preview for this file type.

View file

CRA Git | Maintained and supported by SUSTech CRA and CCSE