Signal Processing

Quiz 2

import numpy as np

import matplotlib.pyplot as plt

from elice_utils import EliceUtils

elice_utils = EliceUtils()

def plt_show():

    plt.savefig("fig")

    elice_utils.send_image("fig.png")

   

def do_fft():

    # (a) generate

    Fs = 100                        # sampling frequency 100Hz

    t = np.arange(0, 2+1/Fs, 1/Fs)  # signal duration 0~2 seconds

    w0 = 5                          # dominant frequency in the signal

    x = np.sin(2*np.pi*w0*t)        # x(t)=sin(2pi w0 t)

    # plot signal

    plt.figure(1)

    plt.plot(t, x)

    plt.grid(True)

    plt.xlabel('t')

    plt.ylabel('x(t)=sin(5 pi t)')

    plt_show()

   

    # (b) fft using numpy fft

    X = np.fft.fft(x) #TODO        # X=fft(x)

    freq = np.linspace(0, Fs, len(X)) #TODO     # freqency axis (w-axis)

    '''np.linsapce(시작, 끝, 샘플 수)'''

    # (c) let's see only the half

    X = X[:len(X)//2] #TODO        # X=fft(x)

    freq = freq[:len(freq)//2] #TODO     # freqency axis (w-axis)

    '''절반만 보이는 이유는 대칭이기 때문'''

   

    return freq, X

   

def main():

    freq,X = do_fft()    

    # plot frequency domain

    plt.figure(2)

    plt.plot(freq, np.abs(X))

    plt.grid(True)

    plt_show()

if __name__ == "__main__":

    main()

 

Quiz 3

from elice_utils import EliceUtils

import numpy as np

import matplotlib.pyplot as plt

import cv2

from answer import convolution_2D

elice_utils = EliceUtils()

def plt_show():

    plt.savefig("fig")

    elice_utils.send_image("fig.png")

def plot_images(images, titles):

    for i, image in enumerate(images):

        'enumerate()는 인덱스와 원소로 이루어진 튜플을 만들어 줌'

        plt.subplot(1, len(images), i+1)

        'subplot(nrows, ncols, index)'

        plt.imshow(image, cmap='gray')

        plt.title(titles[i])

    plt_show()  

def my_convolution_2D(image, kernel):

    # Complete this part for additional bonus points

    pass

   

    return 0 # replace this part with your output

def sobel():

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

    return kernel

def blur():

    kernel = np.array([[1, 2, 1], [2, 4, 2], [1, 2, 1]])

    return kernel

def laplacian():

    kernel = np.array([[0, 1, 0], [1, -4, 1], [0, 1, 0]])

    return kernel

def shadow():

    # Emboss filter

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

    return kernel

def main():

    image1 = cv2.imread("./pikachu.jpg", cv2.IMREAD_GRAYSCALE)

    image2 = cv2.imread("./Lenna.png", cv2.IMREAD_GRAYSCALE)

    result = convolution_2D(image2, sobel())

    '어떤 필터 쓸 건지 선택'

    # Visualizations. Don't touch this line

    plot_images([image2, result], ["Original", "Convolved"])

if __name__ == "__main__":

    main()