Computer Vision Interview 20 essential Q&A Updated 2026
thresholding

Image Thresholding: 20 Essential Q&A

Global vs adaptive methods, Otsu, and when simple thresholding fails.

~10 min read 20 questions Beginner
OtsuadaptivebinaryOpenCV
Quick Navigation
1. What is thresholding?
2. Global threshold
3. Inverse threshold
4. Otsu method
5. Bimodal histogram
6. Adaptive mean
7. Adaptive Gaussian
8. Block size & C
9. Sauvola / Niblack
10. Uneven illumination
11. Thresholding in HSV
12. Multi-level threshold
13. Choosing T
14. Document binarization
15. False positives
16. Morphology after threshold
17. Learned vs classical
18. Soft thresholding (note)
19. uint8 vs float
20. Preprocessing order
1 What is image thresholding? ⚡ easy
Answer: Classifying pixels as foreground vs background by comparing intensity to one or more thresholds—produces binary or multi-label masks.
2 What is global thresholding? ⚡ easy
Answer: Single threshold T for the whole image: pixel → foreground if I > T (or < depending on type). Fast but fails under uneven lighting or overlapping histograms.
3 When use inverse binary threshold? ⚡ easy
Answer: When objects are darker than background (or you want white objects on black mask). Complement of standard binary THRESH_BINARY.
4 How does Otsu choose T? 📊 medium
Answer: Assumes roughly bimodal histogram; picks T that minimizes intra-class variance (equivalently maximizes between-class variance). Automatic, no manual T. 
_, bw = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
5 When does Otsu fail? 📊 medium
Answer: Unimodal or flat histograms, uneven illumination, low contrast, or when foreground fraction is tiny—histogram may not have clear valleys.
6 What is adaptive mean threshold? 📊 medium
Answer: For each pixel, threshold = mean of local neighborhood − C. Handles varying illumination; needs block size larger than foreground features.
7 Adaptive threshold with Gaussian weights? 📊 medium
Answer: Local threshold from Gaussian-weighted mean instead of flat mean—smoother local estimates, slightly better on gradual shading.
8 What is block size in adaptive threshold? ⚡ easy
Answer: Odd window size defining local neighborhood. Too small: noisy mask; too large: loses detail near object boundaries.
9 What is Sauvola / Niblack? 🔥 hard
Answer: Local methods using mean and standard deviation to set threshold—good for document and degraded scans with uneven background.
10 Fix uneven lighting before threshold? ⚡ easy
Answer: Homomorphic filtering, background normalization, CLAHE on luminance, or large-kernel low-pass estimate of illumination to flatten.
11 Threshold colored objects? 📊 medium
Answer: Often convert to HSV and threshold H/S/V ranges (e.g. colored ball)—more robust than RGB for hue-based objects under some lighting.
12 What is multi-level thresholding? 📊 medium
Answer: Several thresholds to get multiple classes (e.g. tissue types). Extension of Otsu exists (multi-Otsu) but cost grows with levels.
13 How pick T without Otsu? 📊 medium
Answer: Manual inspection, ROC on validation set, entropy methods, or trial with domain constraints (known object brightness).
14 Common approach for scanned documents? ⚡ easy
Answer: Adaptive threshold or Sauvola-class; deskew/denoise first; morphology to clean speckles—DL methods also used for hard cases.
15 Why binary masks have holes / noise? ⚡ easy
Answer: Sensor noise, shadows, partial overlap of histograms—use morphology, median blur pre-threshold, or adaptive methods.
16 Apply morphology after thresholding? 📊 medium
Answer: Yes—opening removes pepper noise, closing fills small holes in foreground; preserves label if structuring element smaller than features.
17 Do deep nets replace thresholding? 📊 medium
Answer: For complex scenes, semantic segmentation wins; classical thresholding remains fast for controlled lighting, industrial vision, and documents.
18 What is soft thresholding (wavelets)? 🔥 hard
Answer: Shrinks coefficients toward zero—used in denoising, not classical image binarization. Mention only if interviewer asks denoising context.
19 Threshold on float vs uint8? ⚡ easy
Answer: Same logic but ensure consistent range ([0,1] vs 0–255). Always know your image dtype before comparing to T.
20 Typical order: blur → threshold → morph? 📊 medium
Answer: Often denoise/blur lightly → threshold → morphological cleanup—order depends on whether blur destroys thin structures.

Thresholding Cheat Sheet

Global
  • Single T
  • Otsu if bimodal
Adaptive
  • Local mean/Gauss
  • Block size + C
Docs
  • Sauvola family
  • Illumination first

💡 Pro tip: Say when global fails (lighting) and name one adaptive fix.

Full tutorial track

Go deeper with the matching tutorial chapter and code examples.

Thresholding Tutorial
Previous Next