128 2.
Classification of Types of Defects
Experimental done in testing the effectiveness of the system in detecting glue defects occur in gluing pyramid. 50 tested images being run using this system with 236
defects appeared in the images. The purposed are detecting defects occur in the image and providing location for each defects in x, y and z coordinates.
3. Correction of Defects
Experimental tested by using 15 tested images occur from inspection system. This application used to provide altering the defects in order to reduced rejected product.
Correction done in two steps where 1
st
CoD introduced first time correction and 2
nd
CoD provides additional correction from 1
st
CoD.
5.2 Achievement of Research Objectives
The goal of this research can be summarized as a development of a machine vision system that can be used in automation industry such as inspection of installing windscreen
car. The system’s inspection and control will then determine the results and process of the following steps. The basic requirement of the machine vision system such as object
recognition and 3D transformation in generating x, y and z coordinates are also included. Also the additional steps of ROI creations that substitute the normal method in features
selection in order to cope with unknown objects that were going to be identify later, this method increases flexibility at the cost of accuracy but as long as the images captured are
in fine conditions constant lighting and with a clear and distinct pixel value difference that differs the object from the background the accuracy of the system can achieve up to 98
Experiment 1.
129 In inspection system known as Defect Inspection Monitoring DIM, the
development is based on the characteristic of defect that might occur in the system and saved in system memory. By applying each defect characteristic, the system able to
classify whether the gluing object free from any defects. Once again ROI creation is introduced to train sample defects to be remembered in the system. Each defect recognized
will provides with defect’s information of x, y and z coordinates where the z coordinate generate by multiplying y coordinates with Multiplication Factor MultFactor occur from
Defect Shape Pointing DSP application. DIM Experiment 2 able to achieve 95.76 efficiency that is greater from the other two researchers that develop inspection system for
weld and fabric defect. Development of Correction of Defect CoD is introduced to alter the uncertainties
occur in DIM. The purpose is to reduce number of reject product in order to increase production as well as quality of the product. Information recorded from DIM will sent
automatically to the KUKA arm robot and synchronously alters the reject object according to their characteristic. Reject object occur in DIM application will directly through 1
st
CoD combine with DIM process to check again the gluing object. 2
nd
CoD will take placed when process on 1
st
CoD still occur any defects. The efficiency of this system is quite good where 1
st
CoD recorded 84.75 manage to alter 50 defects out of 59 that tested in this research. 2
nd
CoD complete the system with 100 efficiency manage to alter all the remaining defects occur from 1
st
CoD Experiment 3. This system’s advantages will be mainly on its user define solutions, which the
users can decide and manipulate how they want the system to react to the input that were predetermined. Other than that, the use of an interface bet
ween the system’s vision system PC and KUKA arm robot not only enables the system to be compatible with different
130 brands of arm robot system, it also greatly reduces the set-up time during product changes
as the modifications are towards hardware alteration and not software alterations.
5.3 Significance of Research