绘制匹配关键点时断言失败 BF BRISK+FREAK

Question

我正在尝试使用BRISK+FREAK 拼接两张图片

这是代码，当我尝试绘制匹配项时出现错误

错误：OpenCV(4.1.2) /io/opencv/modules/features2d/src/draw.cpp:225: 错误: (-215: 断言失败) i1 >= 0 && i1

import cv2
import numpy as np
import matplotlib.pyplot as plt

trainImg = cv2.imread('/content/img1.JPG')
trainImg_gray = cv2.cvtColor(trainImg, cv2.COLOR_BGR2GRAY)

queryImg = cv2.imread('/content/img2.JPG')
queryImg_gray = cv2.cvtColor(queryImg, cv2.COLOR_BGR2GRAY)

def detectAndDescribe(image, method=None):
    """
    Compute key points and feature descriptors using an specific method
    """
    
    descriptor = cv2.BRISK_create()
    # get keypoints and descriptors
    (kps, features) = descriptor.detectAndCompute(image, None)

    freakExtractor = cv2.xfeatures2d.FREAK_create()
    keypoints,descriptors= freakExtractor.compute(image,kps)

    return (keypoints, features)

method = 'brisk'
feature_extractor = 'brisk'
feature_matching = 'bf'
kpsA, featuresA = detectAndDescribe(trainImg_gray, method=feature_extractor)
kpsB, featuresB = detectAndDescribe(queryImg_gray, method=feature_extractor)

"Create and return a Matcher Object"
createMatcher = lambda crossCheck :  cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=crossCheck)

def matchKeyPointsBF(featuresA, featuresB, method):
    bf = createMatcher(crossCheck=True)
        
    # Match descriptors.
    best_matches = bf.match(featuresA,featuresB)
    
    # Sort the features in order of distance.
    # The points with small distance (more similarity) are ordered first in the vector
    rawMatches = sorted(best_matches, key = lambda x:x.distance)
    print("Raw matches (Brute force):", len(rawMatches))
    return rawMatches

print("Using: {} feature matcher".format(feature_matching))

fig = plt.figure(figsize=(20,8))

matches = matchKeyPointsBF(featuresA, featuresB, method=feature_extractor)
img3 = cv2.drawMatches(trainImg,kpsA,queryImg,kpsB,matches,None,flags=cv2.DrawMatchesFlags_NOT_DRAW_SINGLE_POINTS)
    

plt.imshow(img3)
plt.show()

这是我得到的完整错误

使用：bf 特征匹配器原始匹配（蛮力）：1967 -------------------------------------------------- -------------------------错误回溯（最近的调用 最后）在（） 4 5 个匹配项 = matchKeyPointsBF(featuresA, featuresB, method=feature_extractor) ----> 6 img3 = cv2.drawMatches(trainImg,kpsA,queryImg,kpsB,matches,None,flags=cv2.DrawMatchesFlags_NOT_DRAW_SINGLE_POINTS) 7 8

错误：OpenCV(4.1.2) /io/opencv/modules/features2d/src/draw.cpp:225： 错误：（-215：断言失败）i1 >= 0 && i1

似乎不知道这里出了什么问题，发现此OpenCV Sift/Surf/Orb : drawMatch function is not working well 无法理解如何纠正此问题

Answer 1

您正在混合来自 FREAK 的 keypoints 和来自 BRISK 的 features：

仔细看detectAndDescribe的代码：

def detectAndDescribe(image, method=None):
    """
    Compute key points and feature descriptors using an specific method
    """
    
    descriptor = cv2.BRISK_create()
    # get keypoints and descriptors
    (kps, features) = descriptor.detectAndCompute(image, None)

    freakExtractor = cv2.xfeatures2d.FREAK_create()
    keypoints,descriptors= freakExtractor.compute(image,kps)

    return (keypoints, features)  # <------ keypoints from freakExtractor.compute and features from descriptor.detectAndCompute

报告的异常看起来是随机的，所以很难找到问题...

---

您可以按如下方式实现detectAndDescribe：

• 用 BRISK 检测关键点
• 将检测到的关键点传递给 FREAK
• 返回freakExtractor.compute的输出

---

建议的实现：

def detectAndDescribe(image, method=None):
    descriptor = cv2.BRISK_create()
    kps = descriptor.detect(image) # kps, features = descriptor.detectAndCompute(image, None)
    freakExtractor = cv2.xfeatures2d.FREAK_create()
    keypoints, descriptors= freakExtractor.compute(image, kps)
    return (keypoints, descriptors)