ArticlePDF Available

Canny Scale Edge Detection

Authors:
R. Pradeep Kumar Reddy
R. Pradeep Kumar Reddy
R. Pradeep Kumar Reddy
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Chiluka Nagaraju at YSR Engineering College of Yogi Vemana University
Chiluka Nagaraju
  • YSR Engineering College of Yogi Vemana University
I Raja Sekhar Reddy at Fourth Paradigm Institute
I Raja Sekhar Reddy
Download full-text PDF
Read full-text
Download citation

Figures

Figures - uploaded by I Raja Sekhar Reddy
Author content
All figure content in this area was uploaded by I Raja Sekhar Reddy
Content may be subject to copyright.
Content uploaded by I Raja Sekhar Reddy
Author content
All content in this area was uploaded by I Raja Sekhar Reddy on Nov 21, 2017
Content may be subject to copyright.
International Journal of Engineering Trends and Technology (IJETT) – Volume X Issue Y- Month 2015
ISSN: 2231-5381http://www.ijettjournal.org Page 1
CANNY SCALE EDGE DETECTION
R. Pradeep Kumar Reddy#1, Dr. C. Nagaraju*2, I. Rajasekhar Reddy#3
#1Assistant Professor, Department of CSE
Y.S.R.Engineering College of YV University, Proddatur, Andhra Pradesh, India.
*2Associate Professor, Department of CSE
Y.S.R.Engineering College of YV University, Proddatur, Andhra Pradesh, India.
#3M.Tech Student, Department of CSE
JNTU Pulivendula, Andhra Pradesh, India.
1pradeepmadhavi@yogivemanauniversity.ac.in
2 nagaraju.c@yogivemanauniversity.ac.in
3 iraganenirajasekharreddy@gmail.com
Abstract—Edge detection is an essential step in digital image
processing and image segmentation. Digital image processing
is one of the significant tool in image processing and
computer vision. Edge is nothing but an abrupt change in the
intensity values of the image. Edge is an elementary feature of
image, contains rich information which plays a vital role for
obtaining the characteristics of image and also useful in
object recognition. Edge describes boundaries of actual image
and which is a fundamental problem in image segmentation.
Detecting an edge in image reduces the amount of data and
screens out unwanted information, while keeping the
important properties of an image. Canny edge detection
algorithm is one of the optimal edge detectors. In this paper,
scaled canny edge detector and logical OR operation is used
for detection of edges in the image. It has been shown that the
scaled canny edge detection performs better than all the other
edge detection operators. Canny edge detector gives good
edge detection and good edge localization.
Keywords—Edge, Segmentation, Canny operator, Edge
Detection and Localization.
I. INTRODUCTION
A fundamental task in application of computer vision is the
edge detection[1] since from the early 1970's. Edge detection
is an important pre-processing step for tasks including
segmentation, active contours and object recognition. Edge
detection refers to the process of identifying and locating
sharp discontinuities in an image. The abrupt changes in pixel
intensities characterizes the boundaries[2] of objects in an
image. Classical edge detection methods[3] uses the
convolution on the spatial domain of the image with an
operator. An operator (a 2-D filter) is constructed to be
sensitive to large gradients in the image while returning values
of zero in uniform regions. Large number of edge detection
operators[4] available each operator is designed to be
detection of certain type of edges[3]. Edge orientation, noise
and edge structure determine the choice of edge detection
operator. The geometry of the operator determines a
characteristic direction in which it is most sensitive to edges.
Operators can be used to look for horizontal, vertical or
diagonal edges. In noisy images detection of edge[5] is a
difficult task, since both the edges and noise contain high
frequency content. Reduction of noise results distorted and
blurred images. Edge detectors for noisy images can average
enough data to minimize localized noisy pixels. This results in
a reduced amount of accurate localization of the detected
edges. Due to noise, there are problems of edge localization,
high computational time, missing true edges and false edge
detection will occur.
Many methods are available for detection of edges[1]. Edge
detection operators are grouped into two categories Gradient
based and Laplacian based. Differentiation of image [6] is the
most common edge detection method. Gradient operators use
first-order derivatives of the image by looking for the
maximum and minimum. Laplacian operators[7] searches for
zero crossings in the second-order derivatives of the image to
find edges.
A. Fist-Order Derivative of Edge Detection is computed as
follows:
(1)
Magnitude of the vector is the important quantity in edge
detection, denoted with , Where
(2)
Direction of the gradient vector is another important
quantity of edge. That is,
(3)
at every pixel location partial derivative of ∂f/∂x and
∂f/∂y are computed for calculation of gradient of an image.
B. Second-Order Derivative Edge Detection
The Second-order derivative, the Laplacian of a 2-D
function f(x,y) is computed as follows.
(4)
Smoothing is usually combined with a Laplacian as a
precursor to finding edges via zero crossings. The 2-D
Gaussian function
International Journal of Engineering Trends and Technology (IJETT) – Volume X Issue Y- Month 2015
ISSN: 2231-5381http://www.ijettjournal.org Page 2
(5)
Whereσ is the standard deviation, which determines the
degree of blurring in the image. The Laplacian of h is defined
as
(6)
The above function is also referred to as the Laplacian of
Gaussian (LOG).
(7)
Fig. 1Dimension coordinates of Laplacian of Gaussian (LOG).
After calculating the two-dimensional second-order
derivative of an image, we find the value of a point which is
greater than a specified threshold and one of its neighbours is
less than the negative of the threshold. The property of this
point is called zero-crossing and we can denote it as an edge
point.
After finding the two-dimensional second-order
derivative of an image, next find the value of a point that is
greater than a specified threshold and one of its neighbours is
less than the negative of the threshold. This point is called the
zero-crossing and it is considered an edge point.
II. CANNY EDGE DETECTION ALGORITHM
Edge detection[3], particularly step edge detection is an
important technique to extract structural information and
considerably reduces the amount of data to be processed. John
F. Canny in 1986 developed a Canny Edge detector[8], also
known as optimal detector, Canny edge detector aims to
satisfy the three general criteria of edge detection[9].
1. Low error rate- which means that it detects only
existent edges.
2. Good localization- The difference between real edge
pixels and the detected edge pixels have to be
minimized.
3. Minimal response- the detected edge should marked
only once not many times.
The steps of the Canny edge detection are as follows:
1. Remove any noise using a filter. Gaussian filter[10] is
used for removing the noise. An example of Gaussian
kernel filter of size 5 is
(8)
2. Find the gradient of the image.
a. Gradient along x, and y directions are calculated
using the convolution masks of 3x3 as follows.
(9)
(10)
b. Gradient strength and direction of the edges are
calculated as follows:
(11)
(12)
Here Gx and Gy are gradients along x and y directions.
3. Non-maximum suppression is performed. This step
removes pixels which are not considered as part of an
edge. Only thin lines will remain, these contain pixels
which are considered to be part of an edge.
4. Hysteresis- The final step, two thresholds are used
named as upper and lower.
a. If the gradient value of a pixel is higher than the
upper threshold, then the pixel is considered as
an edge pixel.
b. If the gradient value of a pixel is less than the
lower threshold, then the pixel is rejected.
c. If the gradient value of the pixel is between the
lower and upper thresholds, then the pixel will
be accepted only if it is connected to a pixel that
is above the upper threshold.
III. CANNY SCALE EDGE DETECTION METHOD
In the proposed Canny scale edge detection approach,
image is resized with the help of scaling. Scaling is a
insignificant process which involves a trade-off between
effectiveness, softness and sharpness. Scaling is of two types
standard and automatic. In standard the threshold has been
decided manually. Where as in automatic threshold has been
decided depending on the factors that are influencing the
image. In the proposed method standard threshold has been
used as a parameter for effective edge detection with scale
down process using Canny edge detection method.
At first, the Canny edge detection and scaling of pixels are
used, and then edge linking is performed with the logical
operations, consisting the following steps:
International Journal of Engineering Trends and Technology (IJETT) – Volume X Issue Y- Month 2015
ISSN: 2231-5381http://www.ijettjournal.org Page 3
1) Read the image.
2) Convert RGB image into gray scale image, if it is a RGB
color image.
3) Scale down the pixel values of the image with multiples of
16 starts from 32 to 128.
4) Find the edge image by applying canny edge detector for
each scale.
5) Perform logical OR operation for edge linking.
The above approach will detect the edges accurately which
are not seen from canny edge detection.
Fig. 2.(a): original image; (b): detected edges using Sobel method; (c): detected edges using Prewitt method;
(d): detected edges using Roberts method; (e): detected edges using LOG method; (f): detected edges using Canny method;
(g): detected edges using proposed method.
(a)
(b)
(c)
(d)
(e)
(f)
(g)
International Journal of Engineering Trends and Technology (IJETT) – Volume X Issue Y- Month 2015
ISSN: 2231-5381http://www.ijettjournal.org Page 4
IV. EXPERIMENTAL RESULTS
The proposed method has been simulated using MATLAB.
The input images are considered to be of .bmp, .png, and .tif.
Other types of images may also be considered. The precision
is assumed to be 8 i.e. the number of bits per pixel is
8.Different sizes of image files have been tested using various
edge detection methods. Comparison of edge maps, obtained
by a various edge detectors is shown in figure 2. The proposed
method is giving the better result comparison with Sobel,
Roberts, Prewitt, LOG, and Canny edge detection operators.
V. CONCLUSION
Edge detection is an important task in image segmentation
for object recognition. Edge contains elementary features of
an image and also describe boundary of the image. Canny
Edge detector is the optimized edge detector and it aims to
satisfy the general criteria of edge detection. Proposed method
uses both Canny edge detector and scale down method for
edge detection. The results shows that edges are accurately
identified in the proposed method and showing more edges
which are not seen from Canny edge detection. It can be
further extended with automatic scaling for all kinds of
images. The thick edges needs to be sharpened to achieve
better edge detection.
REFERENCES
[1] E. Argyle. ―Techniques for edge detection,‖ Proc. IEEE, vol. 59, pp.
285-286, 1971.
[2] Werner Frei, and Chung-Ching Chen. "Fast Boundary Detection: A
Generalization and a New Algorithm", IEEE Transactions On
Computers, Vol. C-26, No. 10, October 1977.
[3] D. Ziou, S. Tabbone, "Edge Detection Techniques – An Overview",
Pattern Recognition & Image Analysis, 1998, 8 (4), pp.537-559.
[4] Ravi.S, A.M. Khan, "Operators Used In Edge Detection Computation:
A Case Study", International International Journal of Applied
Engineering Research, ISSN 0973-4562 Vol.7 No.11 (2012).
[5] J.W. Modestino, R.W. Fries, "Edge detection in noisy images using
recursive digital filtering", Computer Graphics and Image Processing
Volume 6, Issue 5, October 1977, Pages 409–433.
[6] Armando J. Pinho and Lu´ıs B. Almeida, "A review on edge detection
based on filtering and differentiation", Revista Do Detua, VOL. 2, No1,
1997, pp. 113-126.
[7] V. Berzins, ―Accuracy of laplacian edge-detectors,‖ Comput. Vis.
Graph. Image Process. vol. 27, pp. 195–210, 1984.
[8] J. Canny, ―A computational approach to edge detection,‖ IEEE Trans.
Pattern Anal. Machine Intell., vol. PAMI-8, pp. 679–698, June 1986.
[9] M. Bennamoun, B. Boashash, and J. Koo, ―Optimal parameters for
edge detection,‖ in Proc. IEEE Int. Conf. SMC, vol. 2, 1995, pp. 1482–
1488.
[10] Ito, K. KaiqiXiong, "Gaussian filters for nonlinear filtering problems",
Automatic Control, IEEE Transactions on Volume:45 , Issue: 5 ISSN :
0018-9286, Pages 910 - 927.
... The proposed auto-annotation scheme uses edge detection and algorithms to extract the fire object from photos. First, a fire region mask FRM is constructed, then canny edge detection [16] is used to identify fire object edges. This specifies the output shape's upper and lower angles. ...
... Then, edge detection is applied. Implementing Canny edge detection by; Noise Reduction, Finding the Intensity Gradient of the Image, and Non-maximum Suppression [16]. ...
Wildfire Detection System using YOLOv5 Deep Learning Model
Preprint
Full-text available
  • Aug 2023
Wildfires pose a significant threat to the natural environment as well as public safety.Forest fire detection is critical for effective fighting, as once a wildfire has grown to a certain size, it is difficult to control. Recently, there has been a growing demand for forest areas to install a rapid response system to allow for prompt and timely action in the event of forest fires expanding across large areas. In this paper, a proposed framework for wildfire detection in a video sequence using the YOLOv5 deep learning model is presented and implemented. The interested regions represented by (fire object) in the video sequence are extracted using a new auto-annotation scheme to determine the ROI (Region of Interest) based on the edge detection process. Since the public wildfire datasets are yet confined, therefore we have constructed a new wildfire dataset named WILDFIRE-I dataset composed of variant fire images to conduct the performance evaluation of the proposed system. A comparison study with state of art research was performed in our experiments to demonstrate the efficiency of the proposed system based on common performance evaluation metrics. The experimental results exhibited detection accuracy of fire events close to 98 %, with a manual annotation process, while the proposed annotation process has achieved an accuracy of 96 %, with minimum time processing required for dataset image labelling.
View
... Canny edge detection is a robust technique for identifying edges in images, commonly used in computer vision applications for object detection and recognition [69]. The process involves several steps, including noise reduction, gradient computation, non-maximum suppression, and hysteresis thresholding. ...
Enhancing leaf disease detection accuracy through synergistic integration of deep transfer learning and multimodal techniques
Article
Full-text available
  • Sep 2024
The agricultural sector, a cornerstone of economies worldwide, faces significant challenges due to plant diseases, which severely affect crop yield and quality. Early and accurate detection of these diseases is crucial for effective mitigation strategies. The current methods used often lack accuracy and adaptability, especially in diverse environmental conditions. This study introduces a novel, synergistic approach that integrates deep transfer learning with multimodal techniques, specifically canny edges, colour spectrum intensity analysis, and custom data augmentation strategies. Unlike existing methods that rely solely on pre-trained models, the approach utilised in this study offers an innovative fusion of distinct feature extraction techniques. The canny edges highlighted the structural intricacies of leaf diseases, while colour spectrum intensity analysis enhanced the detection of disease-specific colour markers. The customized data augmentation techniques employed (in the study) was shown to enhance the learning process of the models, resulting in their adaptability to diverse agricultural environments. This integration applied to DenseNet201 and EfficientNetB3, achieved detection accuracies of 99.03 % and 98.23 %, respectively, surpassing traditional models and setting new benchmarks in plant disease detection. These results demonstrate the effectiveness of the proposed multi-faceted approach and its potential to significantly enhance crop disease management systems.
View
... Canny is a multi-step edge detection algorithm developed by John F. Canny in 2001 [13], which proceeds through the following steps: ...
View
... In this research, segmentation process was carried out in several stages, namely median blur filter [6], thresholding and canny filter [7]. Initial data processing aims to process the image for obtaining the characteristics of the image so as to gain more accurate information. ...
Iris Recognition: An analysis using LBP, SVC and Distance metrics
Preprint
Full-text available
  • Dec 2023
In this paper, experiment has been conducted to find the optimum iris recognition system between the combinations Local Binary Pattern (LBP) and Distance metric, and LBP and Linear Support Vector Classifier (SVC). First, experiment has been conducted using LBP and different distance metrics. For each of the distance metric, the FAR, FRR and accuracy have been calculated for different threshold values. From the obtained result, it has been found that cityblock distance gives better accuracy compared to remaining distance metrics and the accuracy obtained is 65.93% on CASIA iris dataset. Secondly, iris recognition has been carried out using Local Binary Pattern (LBP) and Linear Support Vector Classifier (SVC). The combination of LBP and Linear SVC is giving an accuracy of 91.83% on CASIA iris dataset.
View
... The Canny edge detection procedure is as follows [52]: ...
An Edge Computing Infrastructure and Deep Learning Methods for Wildfire Monitoring and Anomalies Detection
Thesis
  • Aug 2023
A wildfire is a fire that is burning out of control in the wild environment, most typically in the rural. Both natural causes, like lightning, and human-caused sparks can ignite forests fire. However, the severity of a wildfire is frequently dependent on the weather. Wind, high temperatures, and a lack of rainfall can dry out trees, bushes, fallen leaves, and branches, creating ideal conditions for a wildfire. The topography also has a significant role, as fires spread more rapidly uphill than they do downhill. Recently, an increasing demand is growing for installing a rapid response system in forest regions to enable an immediate and appropriate response to wildfires before they spread across vast areas. This thesis introduces a multilevel system for early wildfire detection to support public authorities to immediately specify and attend to emergency demands. The presented work is designed and implemented within Edge Computing Infrastructure. At beginning, the dataset samples of wildfire represented by a set of video sequences and images are collected and labelled using two methods (Manually and using a new Wildfire auto-annotation scheme to determine the Region of Interest - ROI based on the edge detection process) for training mode purposes. Then, YOLOv5 deep learning model is adopted in the proposed system to build a trained model for distinguishing the fire event against non-fire events in term of binary classification. The proposed system structure comprises Internet of Things (IoT) entities provided with camera sensor capabilities and NVIDIA Jetson Nano Developer kit as an edge computing environment. At the first level, a video camera is employed to assemble environment information received by the micro-controller middle level to handle and detect the possible fire event presenting in the interested area of the acquired images/frames. The last level is characterized as making a decision by sending a text message (alarm) and snapshots (images) of fires event to the cloud server. Meanwhile, a set of commands are sent to IoT nodes to operate the speakers and sprinklers, which are strategically assumed to place on the ground to give an alarm and prevent wildlife loss. The proposed system was tested and evaluated using a wildfire dataset constructed by our efforts. The experimental results exhibited 98% accurate detection of fire events in the video sequence.
View
... These two features are sufficient enough to classify smooth and rough road surfaces. After that individual road surfaces are identified with the help of basic features like edge [13][14][15], histogram [16][17][18] etc. ...
View
Open Signals and Systems Laboratory Exercises: Second Edition
Book
Full-text available
  • Jan 2024
Open Signals and Systems Laboratory Exercises is a collection of lab assignments that have been used in EE 224: Signals and Systems I in the Department of Electrical and Computer Engineering at Iowa State University. These lab exercises have been curated, edited, and presented in a consistent format to improve student learning. This second edition provides a thorough coverage of the MATLAB concepts needed to complete each exercise, and incorporates hardware descriptions and explanations into the exercises that had integrated new lab equipment.
View
Enhancing Lane Detection for Autonomous Vehicles Using Image Processing and Real-Time Analysis
Conference Paper
Full-text available
  • Dec 2023
Lane detection is an essential function for autonomous driving systems since it provides information about the vehicle's position and the road's geometry. However, there are problems with existing lane detection algorithms, such as shadows, uneven lighting, and traffic signs. In this work, we introduce a conventional image processing-based lane detection method based on the Canny edge detector and Hough transform. We test the algorithm on two different platforms, the Raspberry Pi 4 and the Nvidia Jetson Nano boards, utilizing both recorded and live-streaming videos. Our evaluation shows that the technique works effectively for lane detection, with Nvidia Jetson Nano outperforming Raspberry Pi 4. Our method can be used for many autonomous driving tasks, such as lane changing, lane keeping, and collision avoidance.
View
View
Scene Classification using Localized Histogram of Oriented Gradients Method
Article
  • Jan 2016
Scene classification is an important and elementary problem in image understanding. It deals with large number of scenes in order to discover the common structure shared by all the scenes in a class. It is used in medical science (X-Ray, ECG and Endoscopy etc), criminal detection, gender classification, skin classification, facial image classification, generating weather information from satellite image; identify vegetation types, anthropogenic structures, mineral resources, or transient changes in any of these properties. In this paper, at first we propose a feature extraction method named LHOG or Localized HOG. We consider that an image contains some important region which helps to find similarity with same class of images. We generate local information from an image via our proposed LHOG method. Then by combing all the local information we generate the global descriptor using Bag of Feature (BoF) method which is finally used to represent and classify an image accurately and efficiently. In classification purpose, we use Support Vector Machine (SVM) that analyze data and recognize patterns. The basic SVM takes a set of input data and predicts, for each given input, which of two possible classes forms the output. In our paper, we use six different classes of images.
View
View
A review on edge detection based on filtering and differentiation
Article
Full-text available
  • Nov 2011
In this paper we address the issue of edge detec-tion in digital images using linear filtering and differentiation. Our main objective was to join, and in some sense unify, a significant part of the techniques that fall into this paradigm. First, we focus on the question of two-dimensional differenti-ation and the problems associated with its application in the discrete spatial domain. Afterwards, we review some of the most commonly used filtering techniques for edge detection. Resumo – Ao longo deste artigo abordamos a ques ao da detec ao de contornos em imagens digitais, utilizando técnicas baseadas em filtragem e diferencia ao. O nosso principal ob-jectivo foi reunir, e de certa forma unificar, uma parte que julgamos ser significativa dos métodos de detec ao de contor-nos que se enquadram neste paradigma. Em primeiro lugar, começamos por abordar algumas questões relacionadas com a diferencia ao a duas dimensões e também com os problemas que advêm da sua aplica ao no domínio discreto. Em seguida, proporcionamos uma panorâmica de algumas das técnicas ba-seadas em filtragem mais usadas na detec ao de contornos em imagens.
View
Operators used in edge detection computation: A case study
Article
  • Jan 2012
Image segmentation is the processing technique that is helpful to extract the attributes of the image and hence the information. There are various segmentation techniques available depending on the characteristic of pixel considered like intensity value, continuity and degree of similarity between pixels and so on. The edges are one such characteristic of image which is dependent on the discontinuity in intensity value of pixels. The edges are the local change in the intensity values in an image. There are various edge detecting technique that are used in image processing. This paper gives a comparative over-view of the different edge detecting techniques along with the experimental results that are obtained using MatLab software.
View
Edge detection in noisy images using recursive digital filtering
Article
  • Oct 1977
  • Comput Graph Image Process
The problem of detecting edges in noisy digitized images is considered. A stochastic model of edge structure is proposed and the edge detection problem formulated as one of least mean-square spatial filtering. The resulting filtering structures are implemented as two-dimensional recursive digital filters. Computer experiments with this class of edge detectors will be described. The results indicate substantial advantages over conventional edge detectors in the presence of noise. Applications to real-world imagery data are discussed.
View
Fast Boundary Detection: A Generalization and a New Algorithm
Article
  • Nov 1977
We study class of fast algorithms that extract object boundaries from digitized images. A set of orthogonal functions related to distinctive image features is presented, which allows efficient extraction of such boundary elements. The properties of these functions are. used to define new criteria for edge detection and a sequential algorithm is presented. Results indicate considerable improvements over, existing techniques, with a very moderate increase of computational cost.
View
Gaussian filters for nonlinear filtering problems
Article
  • Jun 2000
We develop and analyze real-time and accurate filters for nonlinear filtering problems based on the Gaussian distributions. We present the systematic formulation of Gaussian filters and develop efficient and accurate numerical integration of the optimal filter. We also discuss the mixed Gaussian filters in which the conditional probability density is approximated by the sum of Gaussian distributions. A new update rule of weights for Gaussian sum filters is proposed. Our numerical tests demonstrate that new filters significantly improve the extended Kalman filter with no additional cost, and the new Gaussian sum filter has a nearly optimal performance
View
  • Jun 1986
  • 679-698
  • J Canny
J. Canny, -A computational approach to edge detection,‖ IEEE Trans. Pattern Anal. Machine Intell., vol. PAMI-8, pp. 679-698, June 1986.