Professional Documents
Culture Documents
OBJECTIVE
The aim of this research work is to develop an intelligent computer vision system using fuzzy logic, for the classification (grading) of the whole cashew kernel that can have potential to eliminate inefficient manual grading and inspection approach.
Color Characteristic
Cashew kernels shall be white in color and free from damage. Cashew kernels shall be light brown in color and free from damage. Cashew kernels shall be dark brown in color, it may show deep black spot and free from damage.
Dessert Whole(DW)
WHITE WHOLE
Fig. 1
SCORCHED WHOLE
DESSERT WHOLE
PROPOSED SOLUTION
SYSTEM ARCHITECTURE
System Architecture of proposed System is as
shown below:
In this phase, two different RGB front and top view images of the each cashew kernel under investigation are acquired as shown below:
Preprocessing
PROPOSED SOLUTION
Preprocessing phase smoothes the image using 3x3 average filter to handle the distortion.
Segmentation
PROPOSED SOLUTION
Because of the black background in the image, the histogram of this image is always bimodal as shown in Fig.6. Therefore threshold segmentation technique is used which differentiate the cashew kernel region from background and convert the gray scale image into the binary image as shown in Fig. 7.
Feature Extraction
PROPOSED SOLUTION
Color
Extraction:
Information
To extract color information, RGB image is converted to gray-scale image, then the average intensity of the pixels that belongs to the cashew part in the gray-scale image is determined and this intensity value will be used to classify the cashew kernel based on the color.
Feature Extraction
PROPOSED SOLUTION
(a)
(b)
(c)
Feature Extraction
PROPOSED SOLUTION
Length (L) = L1+L2+ . +Ln-1+Ln Width (W) = W1+W2+ . +Wn-1+Wn n Thickness (T) = T1+T2+ . +Tn-1+Tn n
Because
of the irregular shape of the cashew kernel, for better accuracy in the calculation of these features, sampling strategy has been employed.
Classification
PROPOSED SOLUTION
In order to decide the grade of the whole cashew kernel, in the proposed system two level of classification has been employed.
Color based Classification Weight based classification
Based on the color information, cashew kernel is classified as whether it is white whole, scorched whole or dessert whole.
In second level of classification, Fuzzy Inference System (FIS) is designed to estimate the weight of the cashew kernel.
Final Grading
PROPOSED SOLUTION
Finally, the grade of the
WEIGHT
240 240 240
FINAL GRADE
W240 SW240 DW
CONCLUSION
Inspection
and grading of the whole cashew kernels using computer vision is an alternative to manual and mechanical method which has the potential to automate manual grading practices thus standardizing techniques and eliminating tedious human inspection tasks.
REFERENCES
*1+ Balasubramanian, D. (2001). Physical properties of raw cashew nut. Journal of agricultural Engineering Research, 78(3), 291-297. *2+ B.S. Ogunsina, A.I. Bamgboye (2007), Effects of pre-shelling treatment on the physical properties of cashew nut. *3+ Ghazanfari, A., Moghaddam. (1996). Machine vision classification of pistachio nuts using pattern recognition and neural network. [4] Seyed M.A.Razzavi, M.Maazaherinasab, F.Nickfar, H. Sanacefard, (2007) Physical properties and image analysis of wild pistachio nut. *5+ Brosnan, T., and Sun, D. W. (2004). Improving quality inspection of food products by computer vision, Journal of Food Engineering, 61, 3-16. *6+ A. Ghazanfari, J. Irudayaraj and A. Kusalik, 1996b. Grading pistachio nuts using a neural networks approach. *7+ K. Tanaka, An Introduction to Fuzzy Logic for Practical Applications.Springer, 1997. *8+ Rafael C. Gonzalez and Richard E. Woods, Digital Image Processing, Prentice-Hall International, 2002. [9] MATLAB documentation on Image Processing Toolbox. www.mathworks.com. [10] Kerala State Cashew Development Corporation. www.cashewcorporation.com
Thank You