Online sewing defect monitoring for SNLS machine by Image Processing Technique

 

Dr. Ramesh Babu. V *, Dr. Karunamoorthy. B

Department. of Textile Technology, Department. of Electrical and Electronics Engineering, Kumaraguru College of Technology, Coimbatore-641049, India.

 

ABSTRACT:

Apparels are subjected to visual examination to detect sewing defects after making of the garments which results in higher rejection, time, cost etc. Sewing defects must be detected early i.e during sewing itself and accurately to overcome above quality issues. Apparels are mostly sewn with lock stitch in straight and curve directions, with different colours and stitches per inch. The paper discuss the on line detection of sewing defects occurring during the sewing process. Common defects such as skipped stitch, missed stitch, or loose stitch occurring in lockstitch are detected and marked. Using image processing methods, the proposed work follows the stitch path by capturing digital images of stich lines in lock stitch sewing machine and processed through PYTHON software to detect the sewing defects and subsequently stop the machine during sewing.

 

 

 

INTRODUCTION:

The Indian apparel industry are nowadays focusing on producing quality garments to withstand the high competition from other countries. The term quality of garment refers to the satisfaction of consumers in terms of producing defects free garments. In sewing the common defects occurs such as miss stitch, skip stich and loose stitch etc. These defects are unacceptable and requires stringent quality monitoring process to avoid rejection. In present system the sewing defects are identified once the process complete and the garment inspected at the end of the table. Due to this the rejection percentage is higher. The cost and time required to repair the garment increases tremendously. Further it leads to shipping non-conforming products and meeting the delivery deadlines are becoming challenging.

 

In the past decades, Apparels are subjected to visual examination to detect sewing defects which was resulted in poor quality and higher rejections, cost and time. To overcome these issues very few work has been carried out to monitor seam and stich during sewing itself. A research work monitoring the consumption of sewing thread of class 300 and 400 stiches using an encoder showed that sewing thread consumption is directly related to stitch balance.¹ Another study reported detection of skip stitch by online system using a laser beam reflection method.² Among the electromechanical recognition systems for uneven formed stitches, the counting of thread consumption and tension is presented in.³ All the above methods are indirectly monitoring the sewing defects which are not accurate system of detecting sewing defects. Many researchers did work on fabric defect monitoring by using image processing techniques.^4-14 A study reported that computerized system based on image processing software as an effective alternative for detection of weaving defects.¹⁵ The next another method to inspect and envision defective objects in a patterned fabric image by the texture structures andcartoon.¹⁶ The authors of ¹⁷ offer a result for seams, in this case, the seam pucker assessment employing a feature extraction. A self-organizing map defect is emphasizing, as another major defect caused by material feed or irregular thread tension. A wide-ranging review paper describes the applications of neural networks in the field of fabric to detect the defects.¹⁸ This proposed work develops a system for on line detection of sewing defects occurring during the sewing stage. By capturing digital images of stich lines in lock stitch sewing machine using image processing methods, and processed through PYTHON software to detect the sewing defects and subsequently stop the machine during sewing.

 

MATERIAL AND METHODS:

 

Fig.1 Machine setup

 

The above block diagram describes about the whole process and the components of online sewing defects monitoring system on SNLS machine. Figure 1 shows the schematic diagram of Single needle lock stitch (SNLS) machine. (1) A web camera (5) fixed on the base frame (2) of machine behind the needle. Digital image processing method is designed and developed as given in Fig.2.A CCD camera equipped with a zoom lens was used to capture the stitch images immediately after the sewing under reflected light. All images were processed using histogram equalization to re-assign the brightness to improve the visual appearance.

 

Fig. 2 Image Processing sequence

Hardware setup

Fig. 3 Hardware setup

 

Preprocessing:

The hardware set up shown in Fig. 3, which constitutes of a. Digital camera set up on machine. b. control unit and c. Display unit. Normally, patterned fabric images acquired from a digital camera are embedded with errors like noise, illumination changes and fickle shadows. It would appear with defective objects caused in manufacturing and affect the image quality. The histogram equalization is one of the most well-known methods for contrast enhancement. To increase the pixel values of fabric images by using exploit it. To reduce the bad effects from those errors, a preprocessing step is first conducted for the sampled images.

·      The samples with defects and without defects should be scanned and stored in the camera.(Figure.4)

·      The camera have to be trained with the help of binary conversion of 1’S and 0’S using Histograms:

·      1’s indicate the defective stitches on the fabric and

·      0’s indicates defect free fabrics while stitching.

 

Fig.4. Samples

 

Image decomposition:

The fundamentals of image processing are known as image decomposition. Here the image captured by the camera is decomposed for splitting a fabric image. Here comes to the stage by stage process. It is attaining the picture by executing image decomposition method. The camera captured patterned fabric images are preprocessed.

 

Open computer vision is specifically used for image processing. OpenCV is a package imported to Python. In this way online stitching defect monitoring concept will be implemented. The camera will be placed in a stand. The stitching in the cloth to be examined is continuously monitored by the camera. The image read by the camera will be processed by the OpenCV.

The Image will be read frame by frame from the camera. This image is our input image. This Image will be obtained as BGR image. The image is converted this image to gray. From this gray image the stitching thread is converted to black image, from setting the threshold values of hue, saturation and lightness in an iterative manner. Improper stitching from the black line image is found when there is gap in-between thread line, when there are higher number of black dots and the relay switch also connected with the circuit where it has light when there is proper stitch the light will be in on condition but when there is a defective stitch the light will be off. For image decomposition method, we tag along fabric image to splitting ().

 

                                                                                                                                  

 

In the objective function of model (equ.1),

Total variation norm=                                                                                                                                (2)

It used for recovering piecewise smooth functions

Restoration Discrepancy=                                                                                                                          (3)

 

It approximates the model of the space of oscillating functions introduced by Ngan [11] for fining the texture arrangement.

 

Positive parameter=&                                                                                                                                         (4)

Objective function is balanced by three terms.

and                                                                                                                                                           (5)

It represent the sketch and texture components of,

First-order derivative =                                                                                                                                                       (6)

Divergence Operator  .

 

 

 

RESULTS AND DISCUSSION:

 

Fig.5 Image of defect free fabric

 

Fig.8. The above image includes the frame image of the fabric after stitching and its binary conversion image, inversion of white to black and black to white i.e. the stitching is proper, so the line conversion is straight according its count and so the result shows as proper stitch in the python shell. Here utilize some measurement metrics to specify the efficiencies. It will do by different methods. The below mention measurement are used to find the texture parameters.

 

Precision = (True Negative+True Positive)/ (True Negative+ True Positive + False Positive +False Negative)

False Positive Rate (L) =False Positive/(False Positive+True Negative)

Correct Positive Rate (M) =True Positive/ (True positive+False Negative)

Positive Prediction Value (J) =True Positive/(True Positive+False Positive)

Negative Prediction Value (K)=True Negative/ (True Negative +False Negative)

 

To utilize the L-M graph, which is produced by (L, M), based on the, False positive and correct positive rate. In pixel approach the comparisons take place between and defective manual-labeled images and detected images. The binary images obtained by finding enrichment step. The images of both pixels are labeled as detected and defective.

 

True positive =1:True negative =0

 

The pixel in the detected image is 1.So that of the Defect fabric=0= false positive

It will be the inverted situation is false negative (FN).In the graph the coordinates(L,M), the false positive rate is indicated in left side corner. It is the good classification of false positive rates. This is known as the perfect classification point. The unsystematic estimate line is indicated away from x=y line.

 

Fig. 6 Image of Defective fabrics

 

Fig 9 The above image includes the frame image of the fabric after stitching and its binary conversion image, inversion  of white to black and black to white i.e. the stitching is improper, so the line conversion is improper and the count also differs and so the result shows as improper stitch in the python shell. It gave command to the Controller. So the supply to the machine relay is stopped. Then the machine comes to the idle state.

 

CONCLUSION:

In this article, proposed a novel image decomposition method for SNLS sewn garment assessment which can expertly locate the locations of faulty objects. In this sewn fabric images and to split the defective stitches from the fault free ones. Using an image processing technique, the right stitch formation is checked along the detected paths and the corresponding defects are highlighted. In existing system the sewing defects are identified once the process complete and the garment inspected at the end of the table which leads to rejections and higher cost. Hence on line sewing defect monitoring system controls the sewing defects during the sewing operation by stoppage of sewing machine when the defects are detected.

 

ACKNOWLEDGEMENT:

The authors are thankful to the authorities of Students-Re’ Team and Kumaraguru College of Technology, Coimbatore for the facilities.

 

CONFLICT OF INTEREST:

The authors declare no conflict of interest.

REFERENCES:

1.     H. Carvalho, A. Rocha, J. L. Monteiro, L. F. Silva et al.), “An innovative device for bobbin thread consumption measurement on industrial sewing machines”, IEEE International Conference on industrial Technology, 2004,pp 1034-1039

2.     W. D. Samuel and V. Poojitha, “A simple system for the online detection of SKIP/LOOP stitches in single needle lockstitch sewing machines,” AUTEX Research Journal, 2010.vol. 10, no. 3, pp. 69–72.

3.     T. G. Clapp, L. H. Olson, K. J. Titus, L. Dorrity et al. , “The online inspection of sewn seams”, National Textile Centre Annual Report, November, 1996.pp 173-182.

4.     Michael K. Ng, Henry Y. T. Ngan, Xiaoming Yuan, and Wenxing Zhang, “Patterned Fabric Inspection and Visualization by the Method of Image Decomposition”, IEEE transactions on auto’ science and Engineering, 2014.vol. 11, no. 3, pp. 943 – 947.

5.     H. G. Bu, J. Wang, and X. B. Huang, “Fabric defect detection based on multiple fractal features and support vector data description,” Eng. Appl. Artificial Intelligence, 2009.vol. 22, no. 2, pp. 224–235,

6.     R. C. Gonzalez, R. E. Woods, S. L. Eddins, “Digital Image Processing using MATLAB”, 2005.ISBN 81-297-0515-X,.

7.     U. Farooq, T. King, P. H. Gaskell, and N. Kapur, “Machine vision using image data feedback for fault detection in complex deformable webs,” Trans. Inst. Meas. Control, 2004.vol. 26, no. 2, pp. 119–137.

8.     PejmanNiksaz,” Automatic Traffic Estimation Using Image Processing”, International Journal of Signal Processing, Image Processing and Pattern Recognition Vol. 5, No. 4, December, 2012, pp.167-174.

9.     Raluca Brad, LaviniaBarac, and Remus Brad, “Defect Detection Techniques for Airbag Production Sewing Stages,” Journal of Textiles, vol. 2014, Article ID 738504, 7 pages, 2014. doi:10.1155/2014/738504.

10.   C.H. Chan, Hugh Liu, Thomas Kwan and G. Pang,  “Automation Technology For Fabric Inspection System,” Proc. Conf. Applications of Automation Science and Technology, City University of Hong Kong, 1998.pp. 24-26, Nov.

11.   H. Y. T. Ngan and G. K. H. Pang, “Novel method for patterned fabric inspection using Bollinger bands,” Opt. Eng., 2006.vol. 45, no. 8,

12.   Z. D. Tsai and M. H. Perng, “Defect detection in periodic patterns using a multi-band-pass filter,” Mach. Vision Appl., 2013.vol. 24, pp. 551–565,

13.   C. Kwak, J.A. Ventura, K. Tofang-Sazi, Automated defect inspection and classification of leather fabric, Intelligent Data Analyzation 5 .2001. 355–370.

14.   A. Abouelela, H.M. Abbas, H. Eldeeb, A.A. Wahdan, S.M. Nassar, Automated vision system for localizing structural defects in textile fabrics, Pattern Recognition Lett. 26 .2005. 1435–1443.

15.   K. Yin and W. Yu, “Image processing for the use of garment production detection system,” in Proceedings of the 1st International Congress on Image and Signal Processing (CISP '08), 2008.vol. 3, pp. 349–352.

16.   B. Karunamoorthy.  and D. Somasundaraseswari ,”Automated patterned fabric fault detection using image processing” International Journal of Advanced Research in Computer Engineering and Technology,2015.Vol.4, pp.63-69.

17.   K. L. Mak and W. Li, “Objective evaluation of seam pucker on textiles by using self-organizing map,” IAENG International Journal of Computer Science, 2008.vol. 35, no. 1, pp. 47–54.

18.   C. L. P. Hui, N. S. Fun, and C. Ip, “Review of application of artificial neural networks in textiles and clothing Industries over last decades,” in Artificial Neural Networks Industrial and Control Engineering Applications, K. Suzuki, Ed., INTECH, 2011.

 

 

 

Received on 02.08.2017                             Accepted on 10.10.2017

©A&V Publications all right reserved