(215:Assertion failed) type == CV_32FC1 || type == CV_32FC2 || type == CV_64FC1 || type == CV_64FC2 in function 'dft'

Question

I am trying to correct exposure in images using fourier. This is the error I am facing

  5     padded = np.log(padded + 1) #so we never have log of 0
  6     global complex
  7     complex = cv2.dft(np.float32(padded)/255.0, flags = cv2.DFT_COMPLEX_OUTPUT)
  8     complex = np.fft.fftshift(complex)
  9     img = 20 * np.log(cv2.magnitude(complex[:,:,0], complex[:,:,1]))

  error: OpenCV(4.4.0) /tmp/pip-req-build-njn2fp78/opencv/modules/core/src/dxt.cpp:3335: error: (-215:Assertion failed) type == CV_32FC1 || type == CV_32FC2 || type == CV_64FC1 || type == CV_64FC2 in function 'dft'

My code :

import cv2
import numpy as np
from math import exp, sqrt

image = cv2.imread("2.png")
# grayimg = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
height, width, alpha = image.shape
dft_M = cv2.getOptimalDFTSize(height)
dft_N = cv2.getOptimalDFTSize(width)

#Filter parameters
yh, yl, c, d0, = 0, 0, 0, 0
#User parameters
y_track, d0_track, c_track = 0, 0, 0
complex = 0

def main():
    #copyMakeBorder(src, top, bottom, left, right, borderType[, dst[, value]]) 
    #BORDER_CONSTANT = Pad the image with a constant value (i.e. black or 0)
    padded = cv2.copyMakeBorder(image, 0, dft_M - height, 0, dft_N - width, cv2.BORDER_CONSTANT, 0) 
    padded = np.log(padded + 1) #so we never have log of 0
    global complex
    complex = cv2.dft(np.float32(padded)/255.0, flags = cv2.DFT_COMPLEX_OUTPUT)
    complex = np.fft.fftshift(complex)
    img = 20 * np.log(cv2.magnitude(complex[:,:,0], complex[:,:,1]))

    cv2.namedWindow('Image', cv2.WINDOW_NORMAL)
    cv2.imshow("Image", image)
    cv2.resizeWindow("Image", 400, 400)

    cv2.namedWindow('DFT', cv2.WINDOW_NORMAL)
    cv2.imshow("DFT", np.uint8(img))
    cv2.resizeWindow("DFT", 250, 250)

    cv2.createTrackbar("YL", "Image", y_track, 100, setyl)
    cv2.createTrackbar("YH", "Image", y_track, 100, setyh)
    cv2.createTrackbar("C", "Image", c_track, 100, setc)
    cv2.createTrackbar("D0", "Image", d0_track, 100, setd0)

    cv2.waitKey(0)     
    cv2.destroyAllWindows()

def homomorphic():
    global yh, yl, c, d0, complex
    du = np.zeros(complex.shape, dtype = np.float32)
    #H(u, v)
    for u in range(dft_M):
        for v in range(dft_N):
            du[u,v] = sqrt((u - dft_M/2.0)*(u - dft_M/2.0) + (v - dft_N/2.0)*(v - dft_N/2.0))

    du2 = cv2.multiply(du,du) / (d0*d0)
    re = np.exp(- c * du2)
    H = (yh - yl) * (1 - re) + yl
    #S(u, v)
    filtered = cv2.mulSpectrums(complex, H, 0)
     #inverse DFT (does the shift back first)
    filtered = np.fft.ifftshift(filtered)
    filtered = cv2.idft(filtered)
    #normalization to be representable 
    filtered = cv2.magnitude(filtered[:, :, 0], filtered[:, :, 1])
    cv2.normalize(filtered, filtered, 0, 1, cv2.NORM_MINMAX)
    #g(x, y) = exp(s(x, y))
    filtered = np.exp(filtered)
    cv2.normalize(filtered, filtered,0, 1, cv2.NORM_MINMAX)

    cv2.namedWindow('homomorphic', cv2.WINDOW_NORMAL)
    cv2.imshow("homomorphic", filtered)
    cv2.resizeWindow("homomorphic", 600, 550)

def setyl(y_track):
    global yl
    yl = y_track
    if yl == 0:
        yl = 1
    if yl > yh:
        yl = yh - 1
    homomorphic()

def setyh(y_track):
    global yh
    yh = y_track
    if yh == 0:
        yh = 1
    if yl > yh:
        yh = yl + 1
    homomorphic()

def setc(c_track):
    global c
    c = c_track/100.0
    if c == 0:
        c_track = 1    
    homomorphic()

def setd0(d0_track):
    global d0
    d0 = d0_track
    if d0 == 0:
        d0 = 1
    homomorphic()

main()

I don't understand the issue I am facing. How can I solve this?

DESCRIPTION

The ideal sizes we can make new borders sizes for the image, on the bottom and right (this can be done anyhow) with constant value. After the padding we can transform to the frequency domain and then make a shift

The homomorphic function is then called by the functions responsible to changing its parameters as the user changes the trackbars. The trackbars were defined at the main function and take a limit, a function related to the bar and a variable corresponding to the actual value. The bars were by pattern maintained in the 0–100 scale and are responsible to change the parameters of the filter.

Answer 1

yep, 3d instead of 2d is the case, here's quick solution:

img = cv2.cvtColor(np.float32(image), cv2.COLOR_BGR2GRAY)
dft = cv2.dft(np.float32(img), flags=cv2.DFT_COMPLEX_OUTPUT)

Answer 2

I think the issue is you have RGB image which has 3 channels. But the DFT is asking for 1 or 2 channels.

So you need to convert RGB to gray.

The assertion is pretty clear I think. (215:Assertion failed) type == CV_32FC1 || type == CV_32FC2 || type == CV_64FC1 || type == CV_64FC2 in function 'dft'

It says that type should be

CV_32FC1: 32-bit float with 1 channel or
CV_32FC2: 32-bit float with 2 channels or
CV_64FC1: 64-bit double with 1 channel or
CV_64FC2: 64-bit double with 2 channels