18340052-何泽-计算机视觉期中作业- Edge_Detection1
需积分: 0 179 浏览量
2022-08-03
13:33:52
上传
评论
收藏 54.14MB PDF 举报
000
001
002
003
004
005
006
007
008
009
010
011
012
013
014
015
016
017
018
019
020
021
022
023
024
025
026
027
028
029
030
031
032
033
034
035
036
037
038
039
040
041
042
043
044
045
046
047
048
049
050
051
052
053
054
055
056
057
058
059
060
061
062
063
064
065
066
067
068
069
070
071
072
073
074
075
076
077
078
079
080
081
082
083
084
085
086
087
088
089
090
091
092
093
094
095
096
097
098
099
100
101
102
103
104
105
106
107
CVPR
#0001
CVPR
#0001
CVPR 2021 Submission #0001. CONFIDENTIAL REVIEW COPY. DO NOT DISTRIBUTE.
Lecture Node for Edge Detection
Anonymous CVPR 2021 submission
Paper ID 0001
Abstract
This article first introduces some concepts related to
edge detection, and then introduces four classic edge detec-
tion operators including sobel, scharr, laplace, and canny
operators, and uses these four operators to process different
pictures and compare them. Then summarized the advan-
tages and disadvantages of these four methods, and finally
proposed the improvement direction for the disadvantages
of the canny operator.
1. Introduction
This part will explain some concepts related to edge de-
tection and some terms needed in the following algorithm,
including image edge, noise, gradient, filtering and so on.
However, this part will only introduce the basic concepts,
and the specific algorithm and operation will be elaborated
in detail in the following part.
1.1. Edge
There is always an edge between two adjacent areas with
different gray values, and the edge is the expression of dis-
continuous gray values.
Edge exists between target, background and region, so it
is the most important basis for image segmentation. Since
the edge is a sign of position and insensitive to the change of
gray level, it is also an important feature of image matching.
1.2. Edge Detection
Edge detection is to find the edge contour of the object
in the image, usually to find the collection of pixel points
in the image whose pixel brightness changes dramatically,
because these collection points often show the rough out-
line of the object. Edge detection can be regarded as a very
important method of image feature extraction in the field of
computer vision. Because if we can detect the outline of the
object, and then carry out certain operations, we can repre-
sent the area, shape and other characteristics of the object.
In practice, the edges we get tend to fall into four cate-
gories:
• Depth discontinuity : the object is in a different plane
of the object
• The discontinuity of the surface direction : such as the
two different faces of the cube
• The object material is different : this will lead to dif-
ferent light reflection coefficient
• Different lighting in the scene : such as the ground
thrown by the tree sprout
1.3. Noise
Image noise refers to the unnecessary or redundant inter-
ference information existing in the image data. All kinds of
factors in the image that hinder people’s acceptance of its
information can be called image noise. Noise in the image
is often shown as some isolated pixel points or pixel blocks
that cause strong visual effect. In general, the noise sig-
nal is not related to the object to be studied. It appears in
the form of useless information and disrupts the observable
information of the image.
For digital image signal, noise is shown as large or small
extreme values. These extreme values act on the real gray
value of image pixels through addition and subtraction,
causing bright and dark point interference to the image,
greatly reducing the image quality, affecting the subsequent
work of image restoration, segmentation, feature extraction,
image recognition and so on.
1.4. Filter
Filtering is the operation of filtering out the frequency of
a particular band in the signal. It is an important measure to
restrain and prevent interference. In the image processing,
filtering is a kind of image preprocessing, filtering in the
image processing signal in the specific band frequency filter,
so as to retain the required band frequency signal, mainly
in the preservation of the details of the image itself as far
as possible under the condition of noise suppression of the
image.
There are two main purposes of filtering:
1
叫我叔叔就行
- 粉丝: 24
- 资源: 323
评论0