如何将Android端ORB生成的关键点JSON转为Python的MatOfKeyPoint

Got it, let's walk through how to turn that JSON string from your Android app into a proper cv2.MatOfKeyPoint object in Python. This relies on OpenCV's Python bindings and the built-in json module—here's a straightforward implementation:

Step 1: Set Up Dependencies #

First, make sure you have the required packages installed:

pip install opencv-python

Step 2: Full Conversion Code #

Here's the complete code that parses the JSON, builds individual cv2.KeyPoint instances, and wraps them into a MatOfKeyPoint:

import json
import cv2

def json_to_mat_of_keypoints(json_str):
    # Parse the JSON string into a list of dictionaries
    keypoint_dicts = json.loads(json_str)
    
    # Create a list to hold cv2.KeyPoint objects
    keypoints_list = []
    for kp_data in keypoint_dicts:
        # Map each JSON field to the cv2.KeyPoint constructor parameters
        # The order matches OpenCV's KeyPoint signature: x, y, size, angle, response, octave, class_id
        kp = cv2.KeyPoint(
            x=kp_data["x"],
            y=kp_data["y"],
            size=kp_data["size"],
            angle=kp_data["angle"],
            response=kp_data["response"],
            octave=kp_data["octave"],
            class_id=kp_data["class_id"]
        )
        keypoints_list.append(kp)
    
    # Convert the list of KeyPoints to MatOfKeyPoint
    mat_kp = cv2.MatOfKeyPoint()
    mat_kp.fromList(keypoints_list)
    
    return mat_kp

# Example usage with your sample JSON
sample_json = '[{"class_id":-1,"x":208.0,"y":2492.0,"size":31.0,"angle":312.0059,"octave":0,"response":1.2840814E-4}, {"class_id":-1,"x":114.0,"y":1898.0,"size":31.0,"angle":312.86102,"octave":0,"response":1.4190769E-4}]'
result_mat = json_to_mat_of_keypoints(sample_json)

# Verify the conversion (optional)
print(f"Number of keypoints in MatOfKeyPoint: {len(result_mat.toList())}")

Key Notes & Edge Cases #

• Field Safety: If your JSON might have missing fields (e.g., some keypoints omit angle), use kp_data.get("angle", -1.0) instead of direct indexing to avoid KeyError exceptions.
• Float Compatibility: Android's scientific notation (like 1.2840814E-4) is automatically parsed to Python floats by the json module—no extra work needed.
• OpenCV Version: Make sure your opencv-python version is compatible (most recent 4.x versions work seamlessly).

内容的提问来源于stack exchange，提问作者Pradeepta Ranjan Choudhury