Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
Evaluation of Vision based Surface Roughness using Wavelet Transforms with Neural Network Approach

Evaluation of Vision based Surface Roughness using Wavelet Transforms with Neural Network Approach

Ratings: (0)|Views: 172 |Likes:
Published by ijcsis
Machine vision for industry has generated a great deal of interest in the technical community over the past several years. Extensive research has been performed on machine vision applications in manufacturing, because it has the advantage of being non-contact and as well faster than the contact methods. Using Machine Vision, it is possible to evaluate and analyze the area of the surface, in which machine vision extracted the information with the help of array of sensors to enable the user to make intelligent decision based on the applications. In this work, Estimation of surface roughness has been done and analyzed using digital images of machined surface obtained by Machine vision system. Features are extracted from the enhanced images in spatial frequency domain using a two dimensional Fourier Transform and Wavelet Transform. An artificial neural network (ANN) is trained using feature extracted values as input obtained from wavelet Transform and tested to get Rt as output. The estimated roughness parameter (Rt) results based on ANN is compared with the Rt values obtained from Stylus method and the best correlation between both the values are determined.
Machine vision for industry has generated a great deal of interest in the technical community over the past several years. Extensive research has been performed on machine vision applications in manufacturing, because it has the advantage of being non-contact and as well faster than the contact methods. Using Machine Vision, it is possible to evaluate and analyze the area of the surface, in which machine vision extracted the information with the help of array of sensors to enable the user to make intelligent decision based on the applications. In this work, Estimation of surface roughness has been done and analyzed using digital images of machined surface obtained by Machine vision system. Features are extracted from the enhanced images in spatial frequency domain using a two dimensional Fourier Transform and Wavelet Transform. An artificial neural network (ANN) is trained using feature extracted values as input obtained from wavelet Transform and tested to get Rt as output. The estimated roughness parameter (Rt) results based on ANN is compared with the Rt values obtained from Stylus method and the best correlation between both the values are determined.

More info:

Published by: ijcsis on Dec 04, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

12/04/2010

pdf

text

original

 
(IJCSIS) International Journal of Computer Science and Information Security,Vol.8, No. 8, November 2010
Evaluation of Vision based Surface Roughness usingWavelet Transforms with Neural Network Approach
*T.K.Thivakaran **Dr.RM.Chandrasekaran
 *Research Scholar, **Professor, Department of CSE,MS University, Annamalai University,Tirunelveli 627012.INDIA Chidambaram 620 024.INDIA
 
 Abstract
---Machine vision for industry has generated a greatdeal of interest in the technical community over the pastseveral years. Extensive research has been performed onmachine vision applications in manufacturing, because it hasthe advantage of being non-contact and as well faster than thecontact methods. Using Machine Vision, it is possible toevaluate and analyze the area of the surface, in which machinevision extracted the information with the help of array of sensors to enable the user to make intelligent decision based onthe applications. In this work, Estimation of surface roughnesshas been done and analyzed using digital images of machinedsurface obtained by Machine vision system. Features areextracted from the enhanced images in spatial frequencydomain using a two dimensional Fourier Transform andWavelet Transform. An artificial neural network (ANN) istrained using feature extracted values as input obtained fromwavelet Transform and tested to get R
t
as output. Theestimated roughness parameter (R
t
) results based on ANN iscompared with the R
t
values obtained from Stylus methodand the best correlation between both the values aredetermined.
 Keywords---
Surface roughness, Machine vision, Milling,Grinding, Wavelet Transform, Neural Network.
I.
 
I
 NTRODUCTION
 The quality of components produced is of main concern tothe manufacturing industry, which normally refers todimensional accuracy, form and surface finish. Therefore,the inspection of surface roughness of the work piece isvery important to assess the quality of a component, whichis normally performed using stylus type devices, whichcorrelate the vertical displacement of a diamond-tippedstylus to the roughness of the surface under investigation.But, the limitations of stylus techniques have already beenreported in detail in [6, 5, 4]. Machine Vision typicallyemploys a camera, a frame grabber, a digitizer and a processor for inspection tasks where precision, repetitionand/or high speed are needed. The histograms of the surfaceimage have been utilized to characterize surface roughnessand quality. Fourier transform (FT) of the digitized surfaceimage in which the magnitude and frequency informationobtained from the FT are used as measurement parametersof the surface finish. These methods use the basicassumption that the surface of the specimen is completelyflat and there is no inclination when the images arecaptured. Even a small inclination of the specimen mayresult in inconsistent estimation of roughness of components using machine vision primarily due to the factthat illumination, shadow on the images is likely to bedifferent.In this work, the machined surfaces are capturedusing a Machine Vision system. Following the imageenhancement, the features are extracted and then theroughness parameters are estimated and analyzed. Herewavelet is used to extract the features of the enhancedimage, and an artificial neural network (ANN) is developedto predict the surface roughness. The results are comparedwith that obtained using the standard stylus method.II.
 
OUGHNESS PARAMETERS
 The machined surfaces are generally characterized by threekinds of errors (i) form errors, (ii) waviness, and (iii)surface roughness. The concept of roughness is oftendescribed with terms such as ‘uneven’,’ irregular’, ‘coarsein texture’, broken by prominences’, and other similar ones(Thomas,1999). Similar to some surface properties such ashardness, the value of surface roughness depends on thescale of measurement. In addition, the concept roughnesshas statistical implications as it considers factors such assample size and sampling interval. The one parameter thatis standardized all over the world and is specified andmeasured far more frequently than any other is thearithmetic average roughness height, or RoughnessAverage. Universally called Ra, it was formerly known asAA (Arithmetic Average) in the United States and CLA(Center Line Average) in the United Kingdom. It is definedas the arithmetic mean of the departures of the profile fromthe mean line.Rq (or also known as RMS) is the root mean-squareaverage of the departures of the roughness profile from themean line. Rq has statistical significance because itrepresents the standard deviation of the profile heights andit is used in the more complex computation of skewness,the measure of the symmetry of a profile about the meanline.… (1)
 
243http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
(IJCSIS) International Journal of Computer Science and Information Security,Vol.8, No. 8, November 2010
 A.
 
 Profiles for Turned Machined Components
Figure 1(a) and figure 1(c) shows the profiles obtainedfor a turned component with a stylus instrument. Similarly,figure 1(b) and 1(d) shows the gap profiles obtained for thesame turned components by diffraction method. In bothgraphs, ‘z’ is the deviation of the points on the profile fromthe mean-line. It can be observed that appreciabledifferences in the diffraction pattern are seen for largevariations in the gap and therefore good comparison of results is guaranteed in both only for turned components of medium roughness. For very rough surfaces scattering isobserved. A limitation in the usage of the different methodsis that the smoothness of the edge plays a crucial role in theevaluation of the finish of the components.Fig. 1(a), (c) The profiles obtained for a turned componentwith a stylus instrument. (b), (d) the gap profiles obtainedfor the same turned components by diffraction method.
 B. Profile for Ground Machined Components
Figure 2(a) shows the profiles obtained for a groundcomponent with a stylus instrument. Similarly, figure 3(b)shows the gap profiles obtained for the same groundcomponents by diffraction method. In both graphs, ‘z’ is thedeviation of the points on the profile from the mean-line.Fig. 2(a) The profile obtained from a ground componentwith a stylus instrument, (b) The gap profile obtained for the same ground component by diffraction method.III.
 
S
PECTRUM TECHNIQUES FOR FEATURE EXTRACTION
 
 A. Fourier spectrum
The Fourier spectrum is the frequency domaincounterpart of the autocorrelation function. The FT of thecorrelation is used, which corresponds to the power spectraldensity function and describe how the power in a signal isdistributed over frequency. The power spectrum can revealthe presence of offset, or periodic structures in a data set.
 B. Wavelet Transform (WT)
The wavelet is a tool in surface texture analysis and candecompose a surface into multi-scale representation in avery efficient way. The wavelet transform (WT) is amapping of the signal to the time-scale joint representation.By WT, the decomposition of a signal with a real
244http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
(IJCSIS) International Journal of Computer Science and Information Security,Vol.8, No. 8, November 2010
orthonormal bases
Ψ
mn
(x) obtained through translation anddilation of a kernel function
Ψ  
(
 x
) known as mother waveletas given in eqn. [2],… (2)Where, m,n are integers. To construct the mother wavelet
Ψ  
(
 x
), it is required to determine a scaling function
φ
(
 x
)given in eqn. [3],… (3)Then, the mother wavelet
Ψ  
(
 x
) is related to the scalingfunction as in eqn. [4],… (4)where,The coefficients
h
(
) have to meet several conditions for theset of basis wavelet functions to be unique, be orthonormaland also have a certain degree of regularity.
C. Wavelet Transform for Signals
In two dimensional cases, the one dimensional wavelettransforms are applied along both the horizontal andvertical directions
()
 x
φ 
is a one dimensional real, sequenceintegral scaling function defined as in [5]
2,
()2(2)
 j j j
 xx
φ φ 
=
… (5)Translation
determines the position of this onedimensional function along the x- axis, scale
 j
determine itswidth along x axis and
2
2
 j
controls its height andamplitude. This one dimensional scaling function satisfiesthese conditions:
 
,
 j
φ 
is orthogonal to its integer translates.
 
The set of functions that can be represented as aseries expansion of 
,
φ 
at low scale is containedwithin those at higher scale.So, the difference between any two sets of 
,
 j
φ 
isrepresented by a companion wavelet function
,
ψ 
definedin eqn. [6],
 
2,
()2(2)
 j j j
 xx
ψ ψ 
=
 … (6)
 
Then, the 2 dimensional DWT functions are thelinear products of scaling and wavelet functions
()
 x
φ 
 and
()
ψ 
yielding the eqn. [7] through eqn. [9].
(,)().()
 H 
 xyxy
ψ ψ φ 
=
… (7)
 
(,)().()
 xyxy
ψ φ ψ 
=
… (8)
(,)().()
 D
 yxy
ψ ψ ψ 
=
… (9)where,
(,)
 H 
 xy
ψ 
,
(,)
 xy
ψ 
and
(,)
 D
 xy
ψ 
are called thehorizontal, vertical and diagonal wavelets. Thus, DWT iswell localized and allows decomposition in three directions,namely, horizontal, vertical and diagonal respectively.
 D. Features of Wavelets
In this application, the features are extracted using awavelet which belongs to a family of orthogonal wavelets.The mother wavelet (DB4), its corresponding scaling andwavelet functions and the decomposition filters are shownin Figure 3 and Figure 4 respectively.Fig. 3 wavelet extractionFig. 4 Decomposition of low-pass filter 
h
 
φ
(-
n
) and high- pass filter 
h
 
ψ
(-
m
)The DB4 scaling function is given by
02121222323
iiiii
ahshshshs
+ + +
= + + +
…(10)
[ ] [ ] [ ] [ ] [ ]
0123
2212223
aihsihsihsihsi
= + + + + + +
… (11)The Daubechies DB4 wavelet function is given by
02121222323
iiiii
cgsgsgsgs
+ + +
= + + +
… (12)
[ ] [ ] [ ] [ ] [ ]
0123
2212223
cigsigsigsigsi
= + + + + + +
… (13)
 
IV.
 
 N
EURAL NETWORKS FOR SURFACE ROUGHNESSASSESSMENT
 The roughness features extracted from the machinedimages, are fed as input to an ANN to predict the roughnessvalue R 
t.
ANN consists of a number of elementary unitscalled neurons. A neuron is a simple processor, which cantake multiple inputs and produce an output. Each input intothe neuron has an associated weight that determines the‘‘intensity’’ of the input. The processes that a neuron performs are: multiplication of each of the inputs by itsrespective weight, adding up the resulting numbers for allthe inputs and determination of the output according to theresult of this summation and an activation function. Data isfed into the network through an input layer, it is processedthrough one or more intermediate hidden layers and finally
245http://sites.google.com/site/ijcsis/ISSN 1947-5500

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->