TEM Journal. Volume 12, Issue 1, pages 111-117, ISSN 2217-8309, DOI: 10.18421/TEM121-15, February 2023.
TEM Journal – Volume 12 / Number 1 / 2023. 111
Comparative Analysis of Image on Several
Edge Detection Techniques
Adi Budi Prasetyo
1
, Rizki Wahyudi
1
, Imam Tahyudin
1
, Selvia Ferdiana Kusuma
2,3
,
Luzi Dwi Oktaviana
1
, Azhari Shouni Barkah
1
, Budi Artono
1
1
Universitas Amikom Purwokerto, Purwokerto, Indonesia
2
Politeknik Elektronika Negeri Surabaya, Surabaya, Indonesia
3
Politeknik Negeri Madiun, Madiun, Indonesia
Abstract – Edge detection of an image is needed to
obtain information related to the size and shape of an
image. There are numerous methods for detecting
edges, including the Prewitt, Laplace, and Kirsch
operators. Each edge detection method has different
performance and results. Therefore, this study aims to
analyze the performance comparison of the Prewitt,
Laplace, and Kirsch operators. The analysis process is
carried out using MSE, PSNR and Image Contrast
values. Based on the experiments that have been
carried out, the best edge detection is produced by the
Prewitt operator. The average MSE and PSNR values
obtained were 4.63 and 41.79 dB. The Laplace operator
is good in the contrast value of 17.77. However, the
contrast value only serves as a supporting parameter to
clarify the differences in the results of each edge
detection operator. So it can be concluded that the
Prewitt edge detection method is the best method
among the other two methods.
Keywords – Edge Detection, Kirsch, Laplace, MSE,
PSNR, Prewitt.
1. Introduction
Image processing is needed to enhanced the quality
of the image to be processed, so that later the image
can be easily interpreted by humans or machines [1],
[2], [3].
DOI: 10.18421/TEM121-15
https://doi.org/10.18421/TEM121-15
Corresponding author: Rizki Wahyudi,
Universi tas Amikom Purwokerto, Purwokerto, Indonesia.
Email: rizkiw@amikompurwokerto.ac.id, rizki.key@gmail.com
Received: 23 September 2022.
Revised: 14 December 2022.
Accepted: 17 January 2023.
Published: 27 February 2023.
© 2023 Adi Budi Prasetyo et al; published
by UIKTEN. This work is licensed under the Creative
Commons Attribution‐ NonCommercial‐NoDerivs 4.0
License.
The article is published with Open Access at
https://www.temjournal.com/
Several types of image processing operations,
such as Image enhancement, image restoration,
image segmentation, image analysis, and image
reconstruction are all subcategories. The image
analysis process is needed to identify parameters
related to the characteristics of objects in the image,
then use these parameters to interpret the image [4],
[5]. Image analysis consists of 3 stages, namely,
feature extraction, segmentation, and classification.
The key factor in feature extraction is the ability in
edge detection.
Edge detection in the image is the process of
generating the edges of objects in the image so that
the boundary line information in the image can be
displayed [6]. The purpose of edge detection is to
improve image detail that contains noise [7]. Edges
of image are considered as important features in
images for estimating the attributes and structure of
edge objects that are usually recognized at the border
between two different image regions [8]. The edge
detection method is divided into seven operators,
namely the Sobel operator, Roberts operator, Laplace
operator, Canny operator, Prewitt operator, Laplacian
of Gaussian (LoG) operator, and Kirsch operator [9].
Improper use of edge detection operators will result
in failed detection [10].
Assessment of edge detection quality can be
done objectively using the calculation of Mean
Squared Error (MSE), Peak Noise to Signal Ratio
(PSNR) and Histogram value. MSE is the average
squared error between the actual value and the
forecast value. PSNR is the ratio of the maximum
value of the bit depth measured by the amount of
influential noise [11]. The histogram is a graph that
describes the spread of pixel intensity values of an
image. Histograms can be used to find out important
information from an image. In addition to objectively
assessing the quality of edge detection, there is also a
subjective quality assessment using the human
senses. However, this is difficult to do because the
assessment is very dependent on human vision [12].