Analysis of Machined Surfaces by Grinding Process Using Machine Vision

ANALYSIS OF MACHINED SURFACES BY GRINDING PROCESS USING MACHINE VISIONa B.S.Prasad a *, M.M.M.Sarcar b, P.M.Valli

Corresponding author: [email protected] Assistant Professor, Mechanical Engg dept., GITAM College of Engg, Visakhapatnam

b Professor, Mechanical Engg dept., College of Engg, Andhra University, Visakhapatnamc Professor, Industrial Production dept.,Colleg of Engg,GITAM,Visakhapatnam.

ABSTRACT:Texture of the machined surface closely related to the cutting tool condition. However the traditional methods for measuring surface roughness normally require direct contact and the measurement takes a long time. In addition, the Ra value measured is along a single line of a cut surface and fails to capture the overall features of machined surface. In this work, the images of the ground specimens are analyzed to extract information and there were related to the wheel condition. The surface images of the specimens are captured using the CCD camera at different number of passes during the grinding process. Then, the captured images are analyzed using the texture analysis methods based on fractal dimension and autocorrelation. Fractal dimension is a useful parameter to characterize roughness in an image. The fractal dimension estimated from the captured image is used successfully to characterize the machined surface. The image texture parameters such as the texture aspect ratio, fastest decay autocorrelation length parameters of 2-D autocorrelation functions are evaluated for the surface of the ground components. The variation of these parameters reveals the changes in the texture of the captured images due to wheel wear and loading. Therefore, the condition of the grinding wheel is correlated with various surface texture parameters and the analysis is presented .Based on this study, it is possible to understand and characterize the condition of the grinding wheel. Using the captured images of components.Keywords: Grinding, Machine Vision, fractal dimension, Autocorrelation, texture aspect ratioINTRODUCTIONComputer-integrated manufacturing requires fast and accurate systems that provide the feedback to control the machining process and improve product quality and productivity. On-line process monitoring has been an active area of research because it is recognized as an essential part of fully automated manufacturing systems. One of the parameters to be controlled in machining is surface finish, which is a vital criterion in the performance and utility of industrial products. The proper functioning of a machined part is in many instances largely dependent on the quality of its surface. Engineering properties such as fatigue, hardness and heat transfer are affected by surface finish. Many methods have been developed to measure and quantify surface roughness. The simplest procedure is a visual comparison with an established standard. Surface-roughness monitoring techniques can be mainly divided into two groups: optical and stylus profiling methods. As a conventional method, stylus profiling is a contact measurement, which is not possible to doing the machining operation and is not suitable for in-process applications. Optical methods overcome many standard problems associated with stylus methods. The major limitation of the optical methods currently being used, such as laser profilemetry, is that they scan the surface line by line and, thus, generate profiles. To characterize completely the surface texture, many measurements have to be taken repeatedly over an area. In manufacturing environments, it is often a challenge to find an effective means of reducing cost and improving product quality. When employed efficiently, tool condition monitoring aids in attaining the above objectives in machining application. In this work the grinding process is considered where the quality of the surface produced is a very important aspect. Related with this many approaches have been proposed to accomplish wheel condition monitoring and some have been successfully employed in industry. Most methods essentially involve processing information such as acoustic emission(AE), vibration signature(acceleration signals),cutting force, etc. even through all these techniques perform reasonably well, the implementation usually requires specially designed equipment that is not suitable or even expensive for industry use.The texture of a ground surface is closely related to the wheel condition. The arithmetic average roughness (Ra) of machined surfaces is found to be highly correlated to the grinding wheel condition. However traditional methods for measuring surface roughness normally require special equipment, and the measurements take a long time. In addition, the Ra value is measured along a single line of a cut surface and fails to capture the overall features of a machined surface. In this work, the machine vision based surface quality inspection system is studied to monitor the grinding wheel condition.

A large number of industrial activities have benefited from the application of machine vision technology. These activities includes delicate electronics component manufacturing, quality textile production, metal product finishing, glass manufacturing, machine parts, printing products, granite quality

1

inspection, integrated circuits(IC) manufacturing and many others. Image processing and machine vision technology improves productivity and quality management and provides a competitive advantage to industries that employ this technology. New software and hardware with more powerful functions are emerging continually in the market. The image processing tool boxes of MATLAB 7.0 offer strong image processing functions, which simplify complex image processing.LITERATURE SURVEY

Over the years, the non-contact optical methods have attracted researcher’s attention for the assessment of surface roughness. Most of the methods are based on statistical measures of grey level images in the spatial domain. Attempts were also made to evaluate the roughness by using models with various combinations of input parameters. The following section gives an overview of the various methods employed in part for obtaining surface roughness of the machined parts.

Several methods for the evaluation of the surface quality of the machined surfaces using machine vision have been developed in the last few years.Lee B.Y et al [10] has proposed the use of computer vision techniques to inspect surface roughness of a work pieces for turning operations. The surface image of the work piece is first acquired using a digital camera and then the feature of the surface image is extracted. A polynomial network using a self organizing adaptive modeling method is applied to construct the relationship between the feature of the surface image and the actual surface roughness.Kiran.M.B et al [8] showed the possibility of quick estimation of the finish of medium rough surfaces. Among the procedures discussed, the direct imaging approach is quick and easy to apply in the shop floor level. This can also identify the surface texture and indicate the process of manufacture. Light sectioning method is fast but requires certain amount of pre-processing before estimating the roughness. Phase shifting approach using grating projection is relatively time consuming and elaborate to be used for shop floor applications could be used for the evaluation of medium roughness using vision approach.E.S Gadelmawla [6] has proposed a new approach for surface roughness characterization using computer vision and image processing techniques. A vision system has been introduced to capture images for surfaces to be characterized and software has been developed to analyze the captured images based on the gray level co-occurrence matrix (GLCM).Nirupam Sarkar and Chaudhuri [13] proposed fractal dimension to characterize roughness using an image. It can be used for texture segmentation, estimation of three- dimensional (3D) shape and obtaining other informations. A new method is proposed to estimate fractal dimension in a two- dimensional (2D) image which can readily be extended to a 3D image as well. The method has been compared with other existing methods which is efficient and accurate.Guangming Zhang and Shiva Kumar, Gopalakrishnan [5] proposed that surface finish of machined parts determines the functionality of the product and also the machining requirements for making the parts. On-line monitoring of surface finish has been an active area of machining research. Conventional contact techniques used for surface-finish measurement are not suitable for in-process measurement as they interfere with the machining operation. In this research, the principle of fractal geometry and image processing techniques are used to implement a practical area-based surface-finish monitoring system. The fractal dimension estimated from the captured image is used successfully to characterize the machined surface.Fan et al [3] proposed a method for online non-contact method for measuring the wear of a form grinding wheel. A CCD (charge couple device)camera with a selected optical lens and a frame grabber was used to capture the image of a grinding wheel. The analog signal of the images were transformed into corresponding digital gray level values. Using the binarisation technique, the images of background and the grinding wheel were segmented. Thus the grinding wheel edge was identified. The ‘mapping function method’ is used to transform an image pixel coordinate to a space coordinate. The signal was sent through an 8255 control card to drive a d.c. motor, and then to control the lens focusing movement to acquire the focal plane. The images before and after the grinding process were captured. The position deviation of the grinding wheel edge was analysed. Then, the grinding wheel wear was evaluated. The wear detection accuracy is about 1μm.S. Kurada et al [9] has proposed image processing techniques to enable direct wear measurement to be accomplished in-cycle. Such a system, characterized by its measurement flexibility, high spatial resolution and good accuracy, is presented in this work. The system consists of a fiber-optic light source to illuminate the tool and a CCD camera (used in conjunction with a high resolution video zoom microscope) to capture

2

the reflected pattern. The extent of the flank wear land has been determined using both textural and gradient operators; a texture operator has been implemented in the final system.The processing scheme devised in this work is superior to previous methods, employing vision, due to the implementation of texture-based segmentation. Parameters, describing the characteristics of the wear profile, are used to describe the wear growth rate. The new measurement system offers a convenient and non-contact method for the assessment of tool wear. The measurement procedure is fast and insensitive to vibration and ambient light variations.Majumdar and Bhushan [12] used the fractal approach to predict the behavior of a surface phenomenon at a particular length scale from the observations at other length scale. Here the fractal geometry is used to correct the multi scale self affine topography by scale independent parameter such as the fractal dimensions that are obtained from the spectra of surface profiles.Bradley. C et al [2], accomplished texture analysis using a machine vision-based sensor system. In particular, an investigation of the condition of a two-flute end mill used in a standard face milling operation is presented. The degree of tool wear is estimated by extracting three parameters from video camera images of the machined surface. The performance of three image-processing algorithms, in estimating the tool condition, is presented: analysis of the intensity histogram; image frequency domain content; and spatial domain surface texture.

Image texture parameters are evaluated and correlated with surface roughness of the component measured using stylus values. As the stylus instrument cannot be used for online measurement, the suitability of machine vision based surface quality assessment system is investigated for grinding assessment system is investigated for grinding process. To control and predict the condition of the grinding wheel.SUMMARY:Machine vision system is characterized by its measurement flexibility, high spatial resolution and good accuracy. Advances in computer vision technology have led to the investigation of its application in surface roughness measurement. Visual information has the advantage, that can be interpreted very easily and due to its high information content is the first choice to investigate typical surface forms, which cannot be extracted from indirect measurement signals. Hence, it will not be long before the non-contact online testing of tools in machining of critical parts whose tolerance finish is to be maintained and it will be done using machine vision system.ESTIMATION OF SURFACE ROUGHNESS

Many methods of surface finish measurement have been developed, ranging from simple touch comparator to sophisticated optical techniques. There are three categories of surface finish measurement methods:(I) Comparison based methods(2) Direct measurement methods(3) Non-contact methods

Extensive research has been performed on optical surface roughness measurement techniques, primarily because of the potential for integrating these in automatic manufacturing systems. These methods have the advantage of being non-contact, therefore non-damaging, and can be used at some distance from the surface being measured. They are faster than the contact methods, and have the capability of measuring surface roughness over an area (three dimensional), and not just a line (two dimensional). Machine vision allows for the assessment of surface roughness without contacting or scratching the surface. It provides the advantages of a measurement process for 100% inspection and the flexibility for measuring the part under test without fixing it in a precise position. Machine vision provides a reliable assessment of surface roughness over a given 2-D area rather than a single line trace in a given time and this makes the estimation method for roughness measurement more reliable.SCOPE OF PRESENT WORK

The work is concerned with the prediction of surface finish of the ground components using a machine vision based system for monitoring the condition of the grinding processes. The experiments were carried out in a surface grinding machine with aluminum oxide grinding wheel and mild steel specimens using same cutting conditions. The images of the specimens were captured after every 20 passes consisting of one forward pass with single spark-out pass. The surface finish of the components is also measured using the stylus instrument to decide about the worn-out stage of the grinding wheel. The fractal dimension and the fastest decay autocorrelation length are used as image texture parameters to understand about the condition of the grinding wheel.

3

MACHINE VISION SYSTEM & IMAGE PROCESSINGIn a machine vision process, information is extracted from visual sensors to enable machines to

make intelligent decisions. Machine vision is many times considered as subset of artificial intelligence. A machine vision system analyses images and produces description of what is imaged. These descriptions must capture the aspect of the objects being imaged that are useful in carrying out some task. Thus the machine vision system is considered elements of a fed back loop that is concerned with sensing, while other elements are dedicated to decision-making and the implementation of these decisions. Machine vision as applied to manufacturing extracts information from visual sensors to make intelligent decisions. Such decisions are needed in quality control (detection of defects), process monitoring (prevention of defects), product routing (parts acquisition and sorting) and statistical reporting (performance evaluation). The three main industrial application categories are inspection, identification, and machine guidance. In many tasks that involve image manipulation, the raw data obtained directly from the sensing or acquisition device will give a relatively poor-quality representation of the object or sense of interest. This problem can arise because the image acquisition device is badly calibrated, or because of the noise in the system, the fact that the image was obtained under poor lighting conditions, the existence of a less than ideal or unfriendly operating environment, erroneous information introduced during data transmission, or for a variety of other reasons. Such poor quality data may often be processed so as to improve its acceptability to an observer or to emphasize certain task-specific feature of interest, or so that it is more suited to some automatic means of interpretation. In order to do this it is generally necessary to transform the image data by some appropriate means, when the transform image may be expected to be of a higher in terms of the criteria of current importance. Two broad categories of operation may be identified in the image processing. In the first category, a prime objective of the image transformation is to improve the visual quality of the raw image as it appears to human observer. Such algorithm may be said to be concerned in a general sense with image enhancement. In the second board group can be found the image transformations concerned primarily with operation intended to modify data in a way that aids interpretation in some more specific way, for example, by normalizing an object within an image according to some topological criteria.

Images were taken using a monochrome 0.5-in format CCD camera fitted with a focal lens with extension tubes to give a magnification of approximately 1.3. Images were captured using a video frame-grabber to a resolution of 768×576 pixels. The tools were supported in a specially designed holder so that they could always be returned to the same place after each period of use on the machine, as well as light from a fibre bundle and incandescent lamp source.

Steps in proposed methodology:

4

EXTRACTION OF FEATURES FROM THE SURFACE IMAGE:TEXTURE ANALYSIS: Image texture, defined as a function of the spatial variation in pixel intensities (gray values), is useful in a variety of applications and has been a subject of intense study by many researchers. It is an important feature for identifying objects or regions of interest in an image [1]. Texture is the most important visual cue in identifying these types of regions. The texture variation of the component produced by the fresh wheel and the worn-out wheel should be identified for process monitoring. In this paper, the analysis of ground specimen images captured at different number of passes is done using the 2-D autocorrelation and fractal dimension was presented.FRACTAL BASED TEXTURE ANALYSIS:Fractal geometry: Fractal geometry, as an extension of classical Euclidean geometry, characterizes the average slope of a profile in the two-dimensional space and reflects the space-filling ability in the three-dimensional space. Fractal parameters have been used to represent accurately the naturally occurring shapes like the profiles of mountains, surface profiles, clouds, leaves, etc., using a simple and compact set of equations [1]. Infinite numbers of fractional dimensions are permitted in fractal geometry as opposed to the three integer dimensions allowed in Euclidean geometry. Fractals are identified by their property of appearing similar to the original image under a range of magnification scales. In broad terms, a fractal is a rough or fragmented geometric shape that can be subdivided in parts, each of which is nearly a reduced copy of the whole. Fractals can be described by a scale invariant parameter called a fractal dimension denoted by D. The fractal dimension is a measure of how densely the fractal occupies the space in which it lies. The basic equation of FD given by:

(1)

In this method, Nr is counted in a different manner, consider the image of size M×M pixel has been scaled down to a size s×s where and s is an integer. Then we have estimate or r =s/M. Now, consider the image as a 3D space with (x, y) denoting 2D position and the third coordinate (z) denoting gray level. The (x, y) space is partitioned in to grids of size s × s. On each grid there is a column of boxes of size s × s × s`. If the total number of gray levels is G then [G/s`] = [M/s]. See for example Fig 5.1 where s=s`=3. Assign numbers 1, 2,...to the boxes as shown. Let the minimum and maximum gray level of the image in (i, j)th grid -gall in box number k and l, respectively. then n r(i,j) = l-k+1 is the contribution of Nr

in (i,j)th grid . For example, in Fig X.1 nr (i, j) = 3-1+1. Taking contributions from all grids, we have:

(2)

Where Nr is counted for different values of r (i.e. different values of s). Then using (X.1) we can estimate D, the fractal dimension, from the least square linear fit of log (Nr) against log (1/r).The reason for counting

5

Nr in this manner is that it gives a better approximation to the boxes intersecting the image intensity surface, which is quantized in space and gray value. This is particularly so when there is sharp gray level variation in neighboring pixels in the image. Box counting in the other methods does not cover the image surface so well and hence cannot capture the fractal dimension for rough textured surface.

Fractal geometry as applied to machined surfaces: An intuitive explanation on how fractal geometry can be used to represent the surface roughness is given. As shown in Fig.X.2, a line which represents an ideal smooth surface profile has a fractal dimension of 1. In general, a surface profile should have a fractal dimension ranging between one and two. Consequently, a filled square is the limit in which the surface profile essentially occupies the entire space and has a fractal dimension equal to 2. Thus, it can be seen that as the surface becomes rougher the fractal dimension increases [1].

Fig:3 Fractal geometry applied to surface finish estimation

AREAL AUTOCORRELATION FUNCTION (AACF):In autocorrelation model a single pixel is considered a texture primitive and primitive tone property is gray level. If the texture primitives are relatively large, the autocorrelation function value decreases slowly with increasing distance while it decreases rapidly if texture consists of small primitives. The autocorrelation function describes the general dependence of the values of the data at one position on the values at another position. It is recognized that the AACF is very useful tool for processing random signals. It provides basic information about the spatial relation and dependence of the data, hence when it is used for surface topographic assessment; it is a good method to indicate randomness and directionality of surface features. To quantify the variation of the autocorrelation function of the component images produced by the fresh wheel and the worn-out wheel, the following parameters are used. The AACF is given by the following equations [7]:

6

Where i=0,1,……m<M; j=o,1…….n<N; τi=iΔx τj=jΔy , n(x, y)is the residual surface after a plane is fit to remove the form (longer wavelength and undulations)from the surface data; m and n are the autocorrelation lengths in the x and y directions, respectively. Statistical confidence decreases as i and j approach M and N, respectively, as fewer data points are available for computation of AACF.This limitation causes x and y to be limited to Tm= ΔxM/2 and Tn= ΔyN/2. The AACF helps to visualize the correlation of surface profile heights in different directions. The decay of correlation will be very slow in the direction of consistent surface height whereas the decay will be very fast in the direction of more random surface height values. If a surface has periodicity, its decay will illustrate that periodicity.Consequently, AACF analysis is used to compute parameters such as texture aspect ratio (Str) and fastest decay autocorrelation length (Sal). The AACF plot for each surface is shown in FigThe Fastest Decay Autocorrelation length (Sal)This is a parameter in length dimension used to describe the autocorrelation character of the AACF.   It is defined as the horizontal distance of the AACF which has the fastest decay to 0.2. In other words the Sal is the shortest autocorrelation length that the AACF decays to 0.2 in any possible direction. It is defined as the horizontal distance of the AACF which has the fastest decay to 0.2 in any possible direction. It is given by

(3)

Where τx and τy are the lag distance in the x and y directions, R (τx, τy) is the autocorrelation functions. A large value of Sal denotes that the surface is dominated by low frequency components. While a small value of the Sal denotes the opposite situation. Texture Aspect ratio of the surface (Str)This is a parameter used to identify texture pattern i.e. uniform texture aspect. This parameter used to define the crest ness of the texture. The AACF of a surface which has a significant long crestness texture has the fastest decay along the perpendicular lay direction and the slowest decay along the lay direction. While the AACF of a surface which has similar texture aspects in all directions decays similarly along all directions. Therefore the texture aspect ratio can be defined as:

(4)

p= the distance that the normalized AACF has the fastest decay to 0.2 in any possible directionq= the distance that the normalized AACF has the slowest decay to 0.2 in any possible direction

The larger values of texture aspect ratio indicate stronger uniform texture aspect in all directions, where as smaller values indicate stronger long crest ness.

Fractal dimension and the fastest decay auto correlation length, texture aspect ratios are calculated and used to identify the changes in the texture of the ground surface specimens images due to wheel wear and loading.

EXPERIMENTAL WORK(PROCEDURE INVOLVED)Experiments were carried out on surface grinding machine using an aluminum oxide grinding

wheel, mild steel specimen, with coolant condition. Grinding wheel specifications and machining parameter are tabulated in Table 1.

Each specimen was subjected to number of passes, 20, 40, 60 up to 300 passes, The images of the specimens were captured at every 20 passes consists of one forward pass with single spark-out pass and images of the specimen were taken at the end of each stage. Captured images are shown in Fig 6.1. Surface roughness was measured using a stylus instrument (Pertho meter). All images are grabbed using a CCD camera (Pulnix TM-6) at 768x565 pixels resolution and then transferred to the PC workstation through a frame grabber.

7

Fig 4. Captured ground surface images at different number of passes (20 to 300 number of passes)

8

SUMMARY:The images of the specimens were captured at every 20 passes consisting of one forward pass with

a single spark-out pass. The surface finishes of the components are also measured using the stylus instrument to decide about the worn-out stage of the grinding wheel. The fractal dimension and the fastest decay autocorrelation length, texture aspect ratio are used as image texture parameters to decide about the condition of the grinding wheel.RESULTS AND DISCUSSIONS

Experiments were carried out on surface grinding machine using, an aluminum oxide grinding wheel, mild steel. Each specimen was subjected to number of passes, 20, 40, 60 up to 300 passes with coolant condition, and the ground surface images were taken at the end of each stage. Surface Roughness values are measured using the stylus instrument. In wet grinding the Specimen 1 corresponds to the new wheel and specimen 15 corresponds to the completely worn out wheel. The grinding conditions and the cutting parameters used for the grinding experiments are shown in Table 6.1.

Fig 5.Captured wheel images

Tab2. Different number of passes and the groundSurface components roughness values using with coolant

9

Fig 6. The variation of Surface roughness parameters with number of passes(a),(b)& (c)

The grinding operation is stable up to 240 passes, after which deterioration in the grinding efficiency was noted. So in grinding towards the end a recovery occurred due to sharpening and unloading of wheel. The finish improved slightly but because of wheel wear, could not achieve starting finish values. The measured surface roughness values Ra, Rq and Rt at different number of passes are shown in Fig.6 (a),(b)and (c).

The stylus instrument cannot be used for online measurement, where a quick assessment of surface quality of the manufactured component is required to monitor the process. Also, a single line samples from a surface cannot be a true representative of the whole surface. Hence, the applicability of the machine vision based surface quality assessment for grinding wheel condition monitoring is presented in this work. The captured images of the components at different number of passes are analyzed using fractal dimension and autocorrelation based texture analysis methods.(Held at NSTL,Viskahaptanam).Surface Roughness prediction through image parameters Fractal dimension (Image parameter)

Fractal dimension(FD) represents an ideal smooth surface profile has a fractal dimension of 1. In general, a surface profile should have a fractal dimension ranging between one and two, for image fractal dimension ranging between two and three.

Tab3. Calculated FD values of the ground specimens images at different number of passes.

10

Consequently, a filled square is the limit in which the surface profile essentially occupies the entire space and has a fractal dimension equal to 2. Thus, it can be seen that as the surface becomes rougher the fractal dimension increases. Calculated fractal dimension values shown in Table 3.AREAL AUTOCORRELATION FUNCTION (AACF)

The shape of the autocorrelation function reveals the efficiency of the grinding in all its aspects and any longer wavelengths in the autocorrelation function of the surface provides information about the end of wheel life [17]. But the visual inspection of the 3-D plots of the aerial autocorrelation function, which is shown in Fig.7.3 & 7.4 for the components produced by the fresh wheel and the worn-out wheel, do not provide any appreciable amount of information. 3-D plot of the Areal autocorrelation coefficients for the component image produced with Fresh wheel(a) and worn-out wheel(b) were presented for better understanding.

Fig.7 (a) Fresh Wheel (b) Worn-out wheelTherefore, to characterize the changes in the autocorrelation function the texture aspect ratio and

the fastest decay autocorrelation length parameters are used in this work. To quantify the variation of the autocorrelation function of the component images produced by the fresh wheel and the worn-out wheel, the following parameters are used. The energy value of the dominant detail channels and the fastest decay auto correlation length, texture aspect ratio from the autocorrelation function are evaluated to characterize the changes in the texture of the ground specimen due to wheel wear and loading.

Fig.8. The variation of the autocorrelation based parameters with the number of passes, (a) Fastest decay auto correlation length (Sal), (b) Texture aspect ratio (Str).

The images of the component captured at the worn-out condition of the wheel Fig 7.1 (b) has higher values, it is due to the presence of deeper grooves and more regions of plastic deformation which will scatter more light. Then, the Texture aspect ratio and the fastest decay autocorrelation length are evaluated from the autocorrelation function of the images captured at different no of passes. The variation of these parameters with the number of passes is shown in Fig.8.(a) and (b).

The Larger value of texture aspect ratio indicates stronger uniform texture aspect in directions where as smaller values indicate long crestness. The images of the component produced by the worn-out wheel are having lesser values for the texture aspect ratio. It is due to the presence of deeper grooves and high roughness which scatters more light. The presence of the autocorrelation function represent the

11

abrasive state of the grinding process, where the dressing of the grinding wheel is required to continue the grinding processes. This parameter is higher for the component produced by the worn-out wheel.CONCLUSIONS

The grinding experiments were carried out in a precision surface grinding machine using aluminum oxide grinding wheel and mild steel specimens. Images of specimens surface produced at each trail are captured at different number of passes with coolant condition. In this work the suitability of machine vision based characterization of ground components is done to analyze the changes in the surface texture due to wheel wear. The surface image texture parameters such as fractal dimension and the texture aspect ratio, fastest decay autocorrelation length of the autocorrelation function are used to identify the changes in the texture of the ground components due to wheel wear. The variations of the parameters reveal the changes in the texture of the finished components due to wheel wear. Hence, the evaluated texture parameter can be used to monitor the condition of the grinding wheel.REFERENCES:

1) Anil. K. Jain, Fundamentals of digital image processing, 1989, prentice Hall information and system sciences series.

2) Bradley. C and Wong Y.S, Surface texture Indicating of Tool Wear- A Machine Vision Approach. Int J Adv Manuf technology (2001) No 17, pp .435-443.

3) Fan. K .C, Lee .M.Z And Mou .J.I, Online On-Contact System For Grinding Wheel Wear Measurement ,advanced manufacturing technology(2002) No 19 , pp. 14-22

4) F Luk, V Hyunh and W North, Measurement of Surface roughness by a machine vision system. Journal of Physics, E Scientific Instrument(1989),No 22, pp. (977-980)

5) Guangming Zhang and Shivakumar, Gopalakrishnan, fractal geometry applied to on-line monitoring of surface finish, International Journal of Machine Tools and Manufacture (1995), Vol 36, No 10, pp. 1137-1150.

6) E.S Gadelmawla, A vision system for surface roughness characterization using the gray level co-occurrence matrix, NDT&E International 37(2004), pp. 577-588.

7) Keer D, Shi F, Brown N, Jackson M, and R.Parkin, Quality inspection of food packaging seals using machine vision with texture analysis, proc. Instn Mech. Engrsn (2004) Vol. 218 part B: J. Engineering manufacture.

8)Kiran .M.B, B. Ramamoorthy and V. Radhakrishnan, “Evaluation of surface roughness by vision system.” International Journal of Machine Tool and Manufacture (1998), Vol 38 (5-6), pp. 685-690.

9) Kurada. S and. Bradley. C, A machine vision system for tool wear assessment, Tribology International (1997) Vol 30, No 4, pp. 295-304.

10) Lee B.Y, Tang Y.S, Surface roughness inspection by computer vision in turning operations, International Journal of Machine Tools and Manufacture 41(2001), pp .1251-1263.

11) Lee Hok Siew, Robert M. Hodegson, Texture Measures for Carpet Wear Assessment, IEEE transactions on Pettern Analysis And Intelligence,Vol 10

No 1(1998), pp. 95-105.12) Manjumdar.A and Bhusan.M, Role of Fractal Geometry in Roughness Characterization and

Contact Mechanics of Surface, Journal of Tribology, (1990), Vol, 112, pp .205-216.13) Niruram Sarkar and B. B. Chaudhari. An efficient approach to estimate fractal dimension of textural

images, pattern Recognition (1992). Vol 25, No 9 , pp. 1035-1041.14) Parkkinen J, Selkainaho K, Oja E, Detecting texture periodicity from the co occurrence matrix,

Pattern Recognition letters 11(1990), pp. 43-50.15) Pfeirfer. T, Wiegers. L, Reliable tool wear monitoring by optimized image and illumination control

in machine vision, Measurement 28 (2000), pp. 209-218.16) Rajnessh Kumar P., Kulashekar, B. Dhanasekar, B. Ramamoorthy, Application of digital image

magnification for surface roughness evaluation using machine vision, International Journal of Machine Tools and Manufacture 45 (2005), pp. 228-234.

17) Robert M. Haralic, K. shanmugam, and Its’Hak Dinstein, Textural Features for Image Classification. IEEE Trans.Syst., Man, Cybern. Vol SMC-3.

18) Whitehouse.D.J, “Review Article- Surface metrology”, Measurement Science and Technology, 1997, Vol.8, pp. (955-972)

12