Wrong binarization of corns' darker parts

Asked Oct 17 '19 at 02:38

Active Oct 17 '19 at 06:36

Viewed 40 times

I'm trying to binarize images of corn kernels using Otsu method, but some parts of the kernels are thresholded incorrectly. I also tried Gaussian blur but it doesn't really help.

Is there anyway to fix this? Thanks.

Here's my code:

img = io.imread("data/corn/IMG_4354.jpg")
img = imutils.resize(img, width=500)

gray = color.rgb2gray(img)
gray = filters.gaussian(gray, sigma=3)
binary = gray > filters.threshold_otsu(gray)

fig, axes = plt.subplots(ncols=3, figsize=(20,20))
axes = axes.ravel()

axes[0].axis('off')
axes[0].imshow(img)
axes[0].set_title("Input", fontsize=20)

axes[1].axis('off')
axes[1].imshow(gray, cmap='gray')
axes[1].set_title('gray', fontsize=20)

axes[2].axis('off')
axes[2].imshow(binary, cmap='gray')
axes[2].set_title('Binarized', fontsize=20)

edited Oct 17 '19 at 06:36

asked Oct 17 '19 at 02:38

minhduc0711

Where is your code? What did you try? Please read this forum's help section on how to ask a proper question and what are appropriate questions. Did you try smoothing or removing noise first? – fmw42 Oct 17 '19 at 02:52
Could you add the original input image to the post? – nathancy Oct 17 '19 at 03:09
@nathancy I added an image but it showed up as a link – minhduc0711 Oct 17 '19 at 03:22

Wrong binarization of corns' darker parts

0 Answers0