Video and Image Processing for Finding Paint Defects using BeagleBone Black

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 07 | July -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 199
Video and Image Processing for Finding Paint Defects using
BeagleBone Black
Mr. Sohan Lokhande1, Mr. P. T. Sasidharan2.
1Student, Electronics Design and Technology, NIELIT, Aurangabad, Maharashtra, India.
2Professor, Dept. of Instrumentation, NIELIT, Aurangabad, Maharashtra, India.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Nowadays Automatic inspection plays an
important role in Industrial Quality Management. This
paper proposes a new computer vision system for automatic
painted car body inspection using Beaglebone Black for
quality control in industrial manufacturing. In most
worldwide automotive industries, the inspection process is
still mainly performed by human vision, and thus, is
insufficient and costly. Therefore, automatic paint defect
inspection is required to reduce the cost and time waste
caused by defects. This new system i.e. BeagleBone Black
(BBB) analyzes the images acquired from car body to detect
different kinds of defects. OpenCV library is used for
performing image processing operations. Initially, defects
are detected using a camera connected to the beaglebone
black and filtered using OpenCV libraries and then localized
by using local binary pattern (LBP) and next, detected
defects are classified into different defect types by using kNN
classifier. A user friendly GUI is made in Qt designer where
the entire video stream, captured images, defect area
highlighted with name of that defect, etc will be displayed
on a remote device. The results show that this method could
detect defects and classify them with high accuracy.
Key Words: BeagleBone Black (BBB), Defect Detection,
kNN classifier, Local binary pattern (LBP), Qt.
1. INTRODUCTION
Quality inspection is an important aspect of modern
industrial manufacturing. In automotive industry,
automatic paint inspection is of crucial importance to
maintain the paint quality. In most worldwide automotive
manufacturers the quality inspection process is still
mainly performed by human vision. However, manual
visual inspection is a qualitative and subjective process
with often unreliable results due to its reliance on
inspectors’ own criteria and experience. Also, it is labor
intensive and time-consuming. Therefore, automatic paint
defect inspection is required to reduce the cost and time
waste caused by defects. BeagleBone Black is a board
specifically developed for image processing. BeagleBone
Black is a minicomputer consisting of an ARM Cortex A8
processor with inbuilt 2GB eMMC, 512MB DDR3, micro
HDMI, USB slot, etc. OpenCV (Open Source Computer
Vision) is a library of programming functions mainly
aimed at real-time computer vision. This library is cross-
platform and free for use under the open-source BSD
license. In this paper a novel approach for automatic car
body paint defect detection and classification using
BeagleBone Black is presented with the use of OpenCV
library. The process is organized into two stages:
1. Image capturing and Defect detection
2. Defect classification
First, a camera is connected to the beaglebone black to its
USB slot for capturing video stream of the car. The video
feed is given to the beaglebone black for further
processing. Images are acquired from the video and then
on the images filtering is done using OpenCV libraries. We
have to import the OpenCV libraries into the python and
change its path. Then joint distribution of local binary
pattern (LBP) operators are used for detecting the
defected areas and in the second stage kNN classifier are
used for predicting type of the detected defect. GUI is
created using Qt creator where all of the above processing
of images, defected images with its classification will be
displayed. Graphical view will make the system more
users friendly.
2. Block Diagram
Fig-1:- Block Diagram
Above shown is a basic block diagram where a camera
collects the images and gives further to the Beaglebone
black for Image Processing. The beaglebone black will
receive the image and first it will perform filtering and
then defect detection over that image and classification of
that defect and all this information will be forwarded to
the monitor where a GUI will display all the information.
2.1. Camera
USB Web camera with any specifications.

2.2. BeagleBone Black
The BeagleBoard is a low-power open-source hardware
single-board computer produced by Texas Instruments.
BBB includes an ARM Cortex-A8 CPU, 512 MB RAM, 2 GB
eMMC flash memory, etc. BBB comes with Angstrom
operating system. The USB camera is connected to the BBB
to the USB slot. Video stream is captured of the car
through the camera and processed by BBB to find the
paint defects if any. Processing involves Image capturing,
defect detection, defect classification. We can change the
operating system of the BBB through booting it with OS
images but we have to download and install every package
we require into it. The OpenCV, python, etc packages must
be installed into the BBB operating system.
2.3. Monitor
The monitor will contain a Graphical User Interface (GUI)
for the end user to operate and find the paint defect on the
images acquired through BBB. GUI will be created using Qt
creator. The GUI will consist of video feed, captured
images, defected images, paint faults classification with its
defect name and a database of the previous images.
3. Image Processing
Image Processing is the analysis and manipulation of a
digitized image, especially in order to improve its quality.
It is a processing of images using mathematical operations
by using any form of the signal processing for which the
input is an image, a series of images, or a video, such as a
photograph or video frame; the output of image
processing may be either an image or a set of
characteristics or parameters related to the image.
Fig-2 Image processing
The Images which are captured from the video will be
provided for further processing. The process for testing
the image will be the same but just a classifier will be used
to detect that particular defect in that image. The classifier
will compare the values of the test image with the values
of the trained dataset. It will take top five matching values
and it will compare its append values. The more Label
values match that defect will be shown.
3.1 Image
When we click the “Start Video” button on the GUI the
video will start and when we press the “Space” button on
the keyboard the image frame will be captured from that
video feed. The captured image will be displayed inside
the “Label 6” of the GUI created in Qt designer. So again
when we start the video and capture the image the latest
captured image will be displayed in Label 6 and the
previous captured image will be displayed inside the
“Label”.
3.2 Gray scale Conversion
The captured image which we be selected will be further
be converted into gray scale image for. Converting the
image makes it easy to find the defect areas due to
uniformity in the image. In beaglebone Black as we are
using the openCV library it contains a function for
converting the image into gray scale. BBB takes the image
into BGR format so the function for converting the image
into gray scale is BGR2GRAY.
3.3 Pre-processing
Pre-processing is a common name for operations with
images at the lowest level of abstraction -- both input and
output are intensity images. Pre-processing is an
improvement of the image data that suppresses unwanted
distortions or enhances some image features important
for further processing. Here we are using the GaussianBlur
function provided in the openCV library for pre-
processing. The Gaussian blur is a type of image-blurring
filters that uses a Gaussian function (which also expresses
the normal distribution in statistics) for calculating
the transformation to apply to each pixel in the image.
3.4 LBP
Local binary patterns (LBP) are a type of visual
descriptor used for classification in computer vision.
The LBP feature vector, in its simplest form, is created in
the following manner:
 Divide the examined window into cells (e.g. 16x16
pixels for each cell).
 For each pixel in a cell, compare the pixel to each of its
8 neighbors (on its left-top, left-middle, left-bottom,
right-top, etc.). Follow the pixels along a circle, i.e.
clockwise or counter-clockwise.
 Where the center pixel's value is greater than the
neighbor's value, write "0". Otherwise, write "1". This
gives an 8-digit binary number (which is usually
converted to decimal for convenience).
 Compute the histogram, over the cell, of the frequency
of each "number" occurring (i.e., each combination of

which pixels are smaller and which are greater than
the center). This histogram can be seen as a 256-
dimensional feature vector.
 Optionally normalize the histogram.
 Concatenate (normalized) histograms of all cells. This
gives a feature vector for the entire window.
In our project we are using 3x3 matrixes so it will
compare the pixel to each of its 8 neighbors (on its left-
top, left-middle, left-bottom, right-top, etc.)
Fig 3- Thresholding using LBP matrix
The output of the LBP will be in a straight form so we have
to transpose the output to make it appear in a straight
form. So we have to define two variables as for
Thresholding and for pixel. Thresholding will consist of
simple loop that if the centre pixel value is greater than or
equal to the side pixel value then just append (1) or if the
value is smaller than the centre pixel then append (0).
3.5 KNN Classifier
In pattern recognition, the k-nearest neighbor
algorithm (k-NN) is a non-parametric method used
for classification and regression. In both cases, the input
consists of the k closest training examples in the feature
space. The output depends on whether k-NN is used for
classification or regression:
 In k-NN classification, the output is a class
membership. An object is classified by a
majority vote of its neighbors, with the object
being assigned to the class most common
among its k nearest neighbors (k is a
positive integer, typically small). If k = 1, then
the object is simply assigned to the class of
that single nearest neighbor.
 In k-NN regression, the output is the property
value for the object. This value is the average
of the values of its k nearest neighbors.
k-NN is a type of instance-based learning, or lazy learning,
where the function is only approximated locally and all
computation is deferred until classification. The k-NN
algorithm is among the simplest of all machine
learning algorithms. The principle behind nearest
neighbor methods is to find a predefined number of
training samples closest in distance to the new point, and
predict the label from these. The number of samples can
be a user-defined constant (k-nearest neighbor learning),
or vary based on the local density of points (radius-based
neighbor learning). The distance can, in general, be any
metric measure: standard Euclidean distance is the most
common choice. The Euclidean distance is calculated by
taking the distance of the test point from the distance of
the measured point
3.6 Detected Defect
The kNN classifier will identify the defect on the basis of
its algorithm the defect name will be displayed on to the
GUI. Text box is present in the GUI named as Detected
Defect where the defect name will be displayed and the
defect will be highlighted inside a bounding box so that we
can identify and find a solution to that defect.
4. GUI
Fig 4- Qt GUI
The GUI is created using PyQt which we have to install into
our BBB. The above GUI has several text box and labels
each have its individual significant importance for
displaying the images and text. When we start the GUI
press the Start Video button so it will activate the camera
connected on the beaglebone black and we will get the
video feed on the screen. Then we press the “Spacebar”
button to capture an image frame from it. Then we can
press the Testing button to start the testing of the
captured image frame. The Image button on the GUI is
given to browse through our previously captured images
and we can select them to solidify our defect detection.
The test results will be displayed inside the text box at the
bottom right corner with the name of the defect displayed
and detected using the design methodology. Again when
we start the video and capture the image frame the

previously captured image will be replaced with the new
image and the previous image will be displayed inside the
label on the top right corner and this process will
continue.
5. Results
The defect is detected and displayed inside the beaglebone
black but we can view it using Qt Designer. The GUI results
displayed are:
Fig 5- BBB Desktop screen
This the Desktop screen of the BeagleBone Black which
will be displayed on the remote computer.
Fig 6- Waterspot defect detected and displayed
The image displayed in the big Label box is the test image
on which we are applying our methodology to find the
Paint defects. The paint defect is detected and displayed in
the detection text box as “Waterspot”
Fig 7-Stonechipping Defect detected and displayed
6. Conclusion
Two-stage method for detection and classification of
painted car body defects has been presented. The
proposed algorithm was effective in detecting various
defects such as Stonechipping, Peeling, Waterspot and
Run, etc. This method is a useful tool for automobile
industries to inspect, localize and classify different defects
in their products. The results presented in section 5, show
that a reliable yet cost-efficient inspection of painted
surfaces is attained matching the needs of industry. All of
the detection is automated using the beaglebone Black.
And the Graphical view makes it easy for the end user to
operate and its own potential
REFERENCES
[1] Parisa Kamani, Elaheh Noursadeghi, Ahmad Afshar,
Farzad Towhidkhah, “Automatic paint detection and
classification”
[2] “Getting started – USB Network and Adapter” Chapter
1. Exploring Beaglebone: Tools and Techniques for
Building with Embedded Linux, Wiley Publications by
Derek Molloy on June 18th 2014.
[3] “Qt on Beaglebone” Chapter 6 Exploring Beaglebone:
Tools and Techniques for Building with Embedded Linux,
Wiley Publications by Derek Molloy on June 18th 2014.
[4] Anand Nayyar and Vikram Puri, “A Comprehensive
Review of the BeagleBone Technology Smart Board
Powered by ARM”, International Journel of Smart Home,
Vol 10, No. 4 (2016) pp. 95-108.
[5] “Programming Qt” by Matthias Kalle Dalheimer
published in 1998.

[6] C. Doring, A. Eichhorn, X. Wang, and R. Kruse,
“Improved Classification of Surface Defects for Quality
Control of Car Body Panels, ” in Proc. IEEE International
Conference on Fuzzy Systems, Canada, 2006, pp. 1476-
1481
[7] “Image processing and openCV” Chapter 11. Exploring
Beaglebone: Tools and Techniques for Building with
Embedded Linux, Wiley Publications by Derek Molloy on
June 18th 2014.
[8] “Learning openCV” a book by Adrian Kaehler and Gary
Rost Bradski, O’reilly publications, published on
September 2008.
[9] R. C. Gonzalez and R. E. Woods, Digital Image
Processing, 2nd ed , New Jersey , Prentice Hall , 2002.
[10] G. M. A. Rahaman and M. M. Hossain, “Automatic
defect detection and classification technique from image: a
special case using ceramic tiles”, IJCSIS, Vol. 1, No. 1, pp.
22-30, May. 2009.

Video and Image Processing for Finding Paint Defects using BeagleBone Black

More Related Content

What's hot (19)

Similar to Video and Image Processing for Finding Paint Defects using BeagleBone Black (20)

More from IRJET Journal (20)

Recently uploaded (20)

Video and Image Processing for Finding Paint Defects using BeagleBone Black