docs/api/matching.mdx

---
title: "imcui.ui.matching"
description: "Image matching and RANSAC filtering functions"
---

# imcui.ui.matching

Core matching functions and RANSAC filtering for robust feature matching.

## Functions

### run_matching

Run image matching with specified matcher.

```python
from imcui.ui import run_matching

result = run_matching(
    image0, image1,
    matcher_name="superpoint-lightglue",
    device="cuda"
)
```

**Parameters:**

| Parameter | Type | Description |
|-----------|------|-------------|
| `image0` | np.ndarray | First image (RGB numpy array) |
| `image1` | np.ndarray | Second image (RGB numpy array) |
| `matcher_name` | str | Name of matcher from vismatch |
| `device` | str | Device: "cuda", "cpu", "mps" |

**Returns:**

Dictionary containing keypoints, matches, and scores.

### run_ransac

Run RANSAC filtering on matched points.

```python
from imcui.ui import run_ransac

filtered_matches = run_ransac(
    kp0, kp1, matches,
    ransac_method="CV2_USAC_MAGSAC",
    threshold=8.0
)
```

**Parameters:**

| Parameter | Type | Description |
|-----------|------|-------------|
| `kp0` | np.ndarray | Keypoints from first image |
| `kp1` | np.ndarray | Keypoints from second image |
| `matches` | np.ndarray | Matched point pairs |
| `ransac_method` | str | RANSAC method for filtering |
| `threshold` | float | RANSAC reprojection threshold |

**Returns:**

Filtered matches after RANSAC filtering.

### filter_matches

Filter matches by confidence threshold.

```python
from imcui.ui import filter_matches

clean_matches = filter_matches(matches, scores, threshold=0.2)
```

**Parameters:**

| Parameter | Type | Description |
|-----------|------|-------------|
| `matches` | np.ndarray | Matched point pairs |
| `scores` | np.ndarray | Confidence scores |
| `threshold` | float | Minimum confidence threshold |

**Returns:**

Filtered matches above threshold.

**Source Code**: [imcui/ui/matching.py](https://github.com/Vincentqyw/image-matching-webui/blob/main/imcui/ui/matching.py)