Opencv_Python中使用AKAZE+GMS算法

最新推荐文章于 2024-08-25 07:57:37 发布

盖世猪猪侠

最新推荐文章于 2024-08-25 07:57:37 发布

阅读量3.7k

点赞数 11

分类专栏： opencv 图像处理文章标签： python

本文链接：https://blog.csdn.net/Lv_MV/article/details/110009419

版权

图像处理同时被 2 个专栏收录

2 篇文章 0 订阅

订阅专栏

opencv

1 篇文章 0 订阅

订阅专栏

该博客介绍了如何利用AKAZE算法进行图像特征检测与描述，然后通过BFMatcher进行匹配，并结合GMS匹配方法优化。同时，展示了多种方式对匹配结果进行可视化，包括线条、点和颜色编码。最后，应用到Oxford自行车图像数据集上，生成了匹配后的 Accent.jpg 图片。

摘要由CSDN通过智能技术生成

import copy
import cv2
import numpy as np
from cv2.xfeatures2d import matchGMS
from enum import Enum

def match(img1, img2):
    AKAZE = cv2.AKAZE_create()
    kp1, des1 = AKAZE.detectAndCompute(img1, None)
    kp2, des2 = AKAZE.detectAndCompute(img2, None)
    matcher = cv2.BFMatcher(cv2.NORM_HAMMING)
    matches_all = matcher.match(des1, des2)
    return matches_all,kp1,kp2


class DrawingType(Enum):
    ONLY_LINES = 1
    LINES_AND_POINTS = 2
    COLOR_CODED_POINTS_X = 3
    COLOR_CODED_POINTS_Y = 4
    COLOR_CODED_POINTS_XpY = 5


def draw_matches(src1, src2, kp1, kp2, matches, drawing_type):
    height = max(src1.shape[0], src2.shape[0])
    width = src1.shape[1] + src2.shape[1]
    output = np.zeros((height, width, 3), dtype=np.uint8)
    output[0:src1.shape[0], 0:src1.shape[1]] = src1
    output[0:src2.shape[0], src1.shape[1]:] = src2[:]

    if drawing_type == DrawingType.ONLY_LINES:
        for i in range(len(matches)):
            left = kp1[matches[i].queryIdx].pt
            right = tuple(sum(x) for x in zip(kp2[matches[i].trainIdx].pt, (src1.shape[1], 0)))
            cv2.line(output, tuple(map(int, left)), tuple(map(int, right)), (0, 255, 255))

    elif drawing_type == DrawingType.LINES_AND_POINTS:
        for i in range(len(matches)):
            left = kp1[matches[i].queryIdx].pt
            right = tuple(sum(x) for x in zip(kp2[matches[i].trainIdx].pt, (src1.shape[1], 0)))
            cv2.line(output, tuple(map(int, left)), tuple(map(int, right)), (255, 0, 0))

        for i in range(len(matches)):
            left = kp1[matches[i].queryIdx].pt
            right = tuple(sum(x) for x in zip(kp2[matches[i].trainIdx].pt, (src1.shape[1], 0)))
            cv2.circle(output, tuple(map(int, left)), 1, (0, 255, 255), 2)
            cv2.circle(output, tuple(map(int, right)), 1, (0, 255, 0), 2)

    elif drawing_type == DrawingType.COLOR_CODED_POINTS_X or drawing_type == DrawingType.COLOR_CODED_POINTS_Y or drawing_type == DrawingType.COLOR_CODED_POINTS_XpY:
        _1_255 = np.expand_dims(np.array(range(0, 256), dtype='uint8'), 1)
        _colormap = cv2.applyColorMap(_1_255, cv2.COLORMAP_HSV)

        for i in range(len(matches)):
            left = kp1[matches[i].queryIdx].pt
            right = tuple(sum(x) for x in zip(kp2[matches[i].trainIdx].pt, (src1.shape[1], 0)))

            if drawing_type == DrawingType.COLOR_CODED_POINTS_X:
                colormap_idx = int(left[0] * 256. / src1.shape[1])  # x-gradient
            if drawing_type == DrawingType.COLOR_CODED_POINTS_Y:
                colormap_idx = int(left[1] * 256. / src1.shape[0])  # y-gradient
            if drawing_type == DrawingType.COLOR_CODED_POINTS_XpY:
                colormap_idx = int((left[0] - src1.shape[1]*.5 + left[1] - src1.shape[0]*.5) * 256. / (src1.shape[0]*.5 + src1.shape[1]*.5))  # manhattan gradient

            color = tuple(map(int, _colormap[colormap_idx, 0, :]))
            cv2.circle(output, tuple(map(int, left)), 1, color, 2)
            cv2.circle(output, tuple(map(int, right)), 1, color, 2)
    return output

#原图
img1_name = 'Oxford_dataset/bikes/img1.ppm'
img2_name = 'Oxford_dataset/bikes/img2.ppm'

img1= cv2.imread(img1_name)
img2 = cv2.imread(img2_name)
matches_all,kp1,kp2 = match(img1, img2)
matches_gms = matchGMS(img1.shape[:2], img2.shape[:2], kp1, kp2, matches_all, withScale=False, withRotation=False,thresholdFactor=6)

output = draw_matches(img1, img2, kp1, kp2, matches_gms, DrawingType.ONLY_LINES)
cv2.imwrite('Accent.jpg', output)
cv2.waitKey(0)