Surface abnormality detection in medical and inspection systems using energy variations in co-occurrence matrixes

2025-05-02 0 0 493.42KB 7 页 10玖币

侵权投诉

Surface abnormality detection in medical and inspection

systems using energy variations in co-occurrence matrixes

Nandara K. Krishnand1, Akshakhi Kumar Pritoonka1, Faeze Kiani2,*

1Department of textural science, Online Computer Vision Research center, India

2Department of electronic science, Online Computer Vision Research center, Iran

*ocvrgroup000@gmail.com

Abstract:

Detection of surface defects is one of the most important issues in the field of image processing and machine vision.

In this article, a method for detecting surface defects based on energy changes in co-occurrence matrices is

presented. The presented method consists of two stages of training and testing. In the training phase, the co-

occurrence matrix operator is first applied on healthy images and then the amount of output energy is calculated. In

the following, according to the changes in the amount of energy, a suitable feature vector is defined, and with the

help of it, a suitable threshold for the health of the images is obtained. Then, in the test phase, with the help of the

calculated quorum, the defective parts are distinguished from the healthy ones. In the results section, the mentioned

method has been applied on stone and ceramic images and its detection accuracy has been calculated and compared

with some previous methods. Among the advantages of the presented method, we can mention high accuracy, low

calculations and compatibility with all types of levels due to the use of the training stage. The proposed approach

can be used in medical applications to detect abnormalities such as diseases. So, the performance is evaluated on 2d-

hela dataset to classify cell phenotypes. The proposed approach provides about 89.56 percent accuracy on 2d-hela.

Key words: surface defect detection; co-occurrence matrix; energy variations; image processing; feature extraction

1. Introduction

The visible surface of any object is called its surface. Therefore, any defect that changes the appearance

of the surface (surface) and creates non-normality is called surface defects. Detection of surface defects is

very important in various fields such as factory production, quality control and medicine. Therefore,

researchers are always trying to design intelligent systems to detect surface defects that can perform this

operation in less time and with higher accuracy. The most famous intelligent defect detection systems,

sometimes called automatic inspection systems, are based on image processing and machine vision

techniques. For example, in [1], a method for detecting wood surface defects is presented. Gash and his

colleagues in [2] proposed a fully automatic method for detecting fabric defects, and similarly, algorithms

for leather [3], agricultural products [4] and metal plates [5] have been mentioned so far.

Most of the methods that have been presented so far are designed only for a specific product or

application (ceramic, paper, leather, etc.) Therefore, in this article, a method for detecting defects is

presented, which can be used for most applications without loss of accuracy by using one training step.

In most of the proposed techniques, they try to analyze the texture of the images first and define

appropriate feature vectors to introduce that texture. Finally, by extracting feature vectors from images

and using classifiers, they do the diagnosis. Therefore, as seen, the most important step in defect detection

is texture analysis and definition of a suitable feature vector that can be used to distinguish different types

of textures from each other.

Tissue analysis techniques [6] can be divided into 4 major groups. Statistical techniques [7], structural [8],

model-based [9] and text-based filter [10]. Co-occurrence matrices are one of the statistical techniques. In

this article, the desired images are first analyzed by co-occurrence matrices in different directions, and

then the energy level of the output images is calculated in different directions. In the following, by

comparing the amount of energy in different states, a suitable feature vector that is a good representative

of the texture of that image is defined. In the following, according to the defining feature vector, a two-

step method for detecting defects is presented. The training phase consists of windowing perfectly intact

images and extracting feature vectors from them. Also, at this stage, by obtaining the average of the

extracted vectors and calculating the distance of each of the windows with the average vector, the quorum

of the windows is obtained. Finally, in the test phase, by windowing the images and comparing with the

quorum, surface defects are revealed and detected. In the results section, to determine the "diagnosis

accuracy" of the presented method, images of surface defects on agricultural products, stone and ceramics

were collected and the method was applied to them. High detection accuracy, lack of dependence on the

application context, and low calculations are among the advantages of the method, which are discussed in

the results section.The proposed approach is a general algorithm which is learned the defect. So, it can be

used in medical applications such as bacteria abnormality detection [15], DNA structure [16], etc.

2. Co-occurrence matrix

Co-occurrence matrices were first proposed by Harlick in [11]. The co-occurrence matrix is a quadratic

operator that examines the spatial relationship of each pixel of the image with its neighbors. The co-

occurrence matrix operator calculates that in the input image, the brightness intensity "j" has been seen

several times after the brightness intensity "i" according to the defined spatial relation. This issue is

shown in figure (1). As seen in Figure (1-A), the input image is 3 levels and the spatial relationship is

defined as the first pixel on the right. Then, the co-occurrence matrix of the input image is calculated in

Figure (1-b). For example, the number 3 in the second row and the first column shows that 3 times the

brightness of zero has occurred in the pixel to the right of the brightness of one.

Figure 1. GLCM process numerical example

a) original image b)produced GLCM

According to the above explanations, the co-occurrence matrix of each image can be calculated according

to different spatial relations. In this regard, generally 8 types of spatial relationships are considered for

calculating the co-occurrence matrix, which are also called 8 directions according to their degree of

rotation with respect to the original pixel. This issue is shown in equation 1.

R=[i,j+1] 0° R=[i-1,j+1] 45° R=[i-1,j] 90° R=[i-1,j-1] 135°

R=[i,j-1] 180° R=[i+1,j-1] 225° R=[i+1,j] 270° R=[i+1,j=1] 315° (1)

After calculating the co-occurrence matrix, the obtained matrix can be shown in the form of an output

image. In figure (2), the image of the co-occurrence matrix calculated for an image of orange travertine

stone texture is shown.

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

Surfaceabnormalitydetectioninmedicalandinspectionsystemsusingenergyvariationsinco-occurrencematrixesNandaraK.Krishnand1,AkshakhiKumarPritoonka1,FaezeKiani2,*1Departmentoftexturalscience,OnlineComputerVisionResearchcenter,India2Departmentofelectronicscience,OnlineComputerVisionResearchcenter,Iran*ocv...

展开>> 收起<<

Surface abnormality detection in medical and inspection systems using energy variations in co-occurrence matrixes.pdf

共7页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Surface abnormality detection in medical and inspection systems using energy variations in co-occurrence matrixes

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: