Skip to content

Find object position based on python-opencv2

License

Notifications You must be signed in to change notification settings

NetEaseGame/aircv

Repository files navigation

aircv

Build Status PyPI

Find object position based on python-opencv2 for python2.7+

Usage

import aircv as ac
imsrc = ac.imread('youimage.png') # 原始图像
imsch = ac.imread('searched.png') # 带查找的部分

SIFT查找图像

print ac.find_sift(imsrc, imsch)
# - when Not found
@return None 
# 之前是返回的 []

# - when found
@return {'point': (203, 245), 'rectangle': [(160, 24), (161, 66), (270, 66), (269, 24)], 'confidence': 0.09}
# point: 查找到的点
# rectangle: 目标图像周围四个点的坐标
# confidence: 查找图片匹配成功的特征点 除以 总的特征点

SIFT多个相同的部分查找

print ac.find_all_sift(imsrc, imsch, maxcnt = 0)
# - when not found
@return []
# - when found
@return [{..}, {..}]
# {..}的内容跟SIFT查找到单个图像的格式一样

maxcnt是可选参数,限制最多匹配的数量。

直接匹配查找图像

print ac.find_template(imsrc, imsch)

期望输出 (目标图片的中心点,相似度), 相似度是电脑计算出来的一个值,跟平常所说的相似97%不是一个意思。对于这个值,达到0.999以上才算是图片一样。

(294, 13), 0.9715

查找多个相同的图片,如在图形

template1

中查找

template2

print ac.find_all_template(imsrc, imsch)

期望输出 (目标图片的中心点,相似度)

[((294, 13), 0.9715), ...]

效果

2res

开发规范

开发规范

LICENSE

LICENCE under MIT

Some other idea. Not implemented

example

import aircv

imsrc = aircv.Image('demo.png')
imobj = aircv.Image('object.png')

print imsrc.find(imobj, method=aircv.FIND_TMPL) # or method=aircv.FIND_SIFT
# expect aircv.Position(x=10, y=20, extra={'method': aircv.FIND\_TMPL, 'result': 0.98})

print imobj.find_in(imsrc, method=aircv.FIND_TMPL)
# expect aircv.Position(x=10, y=20)

rect = aircv.Rect(left=80, top=10, width=50, height=90)
# Rect define: Rect(left=0, top=0, right=None, bottom=None, width='100%', height='100%')
pos = imsrc.find(imobj, rect=rect, method=aircv.FIND_TMPL)
print pos
# expect aircv.Position(x=10, y=20)

print imsrc.draw_point(pos) # .draw_point(pos2)
# expect aircv.Image object

print imsrc.draw_rectangle(aircv.Rect(left=80))
# expect aircv.Image object

print imsrc.draw_circle(??)

print imsrc.cv_object
# expect numpy object

imsrc.save('source.png')
# An Exception raised when file exists

print imsrc.rect() == imobj.rect()
# expect True or False

print imsrc.percent(imobj)

About

Find object position based on python-opencv2

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Contributors 3

  •  
  •  
  •