-215 opencv computation_error ai_generated true

cv::error: (-215:Assertion failed) matches.size() >= 4 in function 'cv::detail::matchesGraphAsString'

ID: opencv/stitching-error-not-enough-matches

Also available as: JSON · Markdown · 中文

80%Fix Rate

85%Confidence

1Evidence

2024-11-05First Seen

Version Compatibility

Version	Status	Introduced	Deprecated	Notes
4.5.5	active	—	—	—
4.8.0	active	—	—	—
4.9.0	active	—	—	—

Root Cause

Image stitching requires at least 4 good matches between images to compute homography, but fewer were found.

generic

中文

图像拼接需要至少4个良好匹配点来计算单应性矩阵，但找到的匹配点少于4个。

Official Documentation

https://docs.opencv.org/4.x/d8/d19/tutorial_stitcher.html

Workarounds

75% success img = cv2.detailEnhance(img, sigma_s=10, sigma_r=0.15) # Or use contrast adjustment: img = cv2.convertScaleAbs(img, alpha=1.5, beta=0)
```
img = cv2.detailEnhance(img, sigma_s=10, sigma_r=0.15)
# Or use contrast adjustment:
img = cv2.convertScaleAbs(img, alpha=1.5, beta=0)
```
80% success sift = cv2.SIFT_create(nfeatures=5000, contrastThreshold=0.03, edgeThreshold=10) kp, des = sift.detectAndCompute(img, None) # Then use FLANN matcher with lower ratio threshold
```
sift = cv2.SIFT_create(nfeatures=5000, contrastThreshold=0.03, edgeThreshold=10)
kp, des = sift.detectAndCompute(img, None)
# Then use FLANN matcher with lower ratio threshold
```

中文步骤

img = cv2.detailEnhance(img, sigma_s=10, sigma_r=0.15)
# Or use contrast adjustment:
img = cv2.convertScaleAbs(img, alpha=1.5, beta=0)

sift = cv2.SIFT_create(nfeatures=5000, contrastThreshold=0.03, edgeThreshold=10)
kp, des = sift.detectAndCompute(img, None)
# Then use FLANN matcher with lower ratio threshold

Dead Ends

Common approaches that don't work:

70% fail
The error is about actual matches found, not the threshold; increasing threshold won't generate more matches.
80% fail
Lower quality images have fewer features, making the problem worse.
50% fail
ORB may produce fewer matches if images have low texture; fine-tuning is needed.