Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/286813400
CITATIONS READS
6 106
5 authors, including:
All content following this page was uploaded by Anjitha Vadukoot on 13 January 2016.
B. Prewitt operator ଵ
ܩሺݔǡ ݕሻ ൌ ቀଶగఙమ ቁ ݁ ିሺ௫
మ ା௬ మ ሻȀଶఙ మ
(3)
Prewitt operator is in the very sense similar to Sobel
edge detector. Also it takes the advantage of 3*3 convolution Where ı represents standard deviation. The smoothened
masks in orthogonal directions. The masks are shown in Figure image is represented by the convolution of original image and
2.
Gaussian function.
• Hysteresis thresholding
Figure. 3. Masks for Roberts operator The process is collectively known as removal of spurious
edges. This method of thresholding makes use of two
thresholds: upper threshold and lower threshold. Thresholds
International Conference on Embedded Systems - (ICES 2014) 9
ோమ
ܴܲܵܰ ൌ ͳͲ݈݃ଵ ሺ ሻ (5)
ெௌா
Where R is the maximum fluctuation in the input image data
type. For a gray level image, R is taken as 255 where each
pixel value is represented as an 8 bit integer.
D. L2RAT
L2RAT is the ratio of the squared norm of the signal or
image approximation to the input signal.
V. RESULTS
The results for various edge detection operators are
obtained as follows. Two different input images are
considered. The edge detected outputs and performance
parameter values for each edge detection operator are
shown.
International Conference on Embedded Systems - (ICES 2014) 10
VI. INFERENCES
By analysing the first input image which has mainly low
frequency contents in it, PSNR value is higher for Sobel
operator and it is lower for Prewitt and the value of the same
for Canny operator is in between of these two. MSE is
minimum for Sobel operator and maximum for Robert operator
and Canny holds a value in between. MAXERR is the same for
all the operators. L2RAT is higher for Sobel operator and
lower for Roberts operator. Canny offers a value in between
them. Canny operator possesses higher processing time than
other operators which is approximately 730ms.
By analysing the second input image which is having high
frequency contents compared to first input image, Roberts
operator offers high PSNR and minimum MSE. Canny
Figure. 5. Comparison of edge detectors (a) input image (b) Sobel (c) operator provides low MAXERR, low L2RAT and high
Prewitt (d) Canny (e) Roberts processing time compared to all other operators.
The two sets of images are compared and by considering all
parameters such as PSNR,MSE,MAXERR,L2RAT canny
operator has the optimal value even though it possess a greater
The following tables give a comparative study of processing time. Ignoring the latter case canny is chosen as the
performance parameters of various edge detection best edge detection operator.
operators.
REFERENCES
[1]. John Canny,”A Computational Approach to Edge
Table 1: Table showing the values for first input image Detection”.IEEE Transactions on Pattern Analysis and
Machine Intelligence, Vol.PAMI-8,NO.6,November 1986.
[2]. Xiaoyang Li, Jie Jiang, Qiaoyun Fan, “An Improved
Real-time Hardware Architecture for Canny Edge
Detection Based on FPGA”.Third International Conference
on Intelligent Control and Information Processing, July 15-
International Conference on Embedded Systems - (ICES 2014) 11