113 The percentage error is too large to be implemented in industry where it can reflect
to their quality product. Without proper step in defining good or poor template will be crucial to machine vision system where it cause system to be less effective and productive.
4.4.2 Constraint in Camera Placement
The important step to vision system is image acquisition, where the captured image of the object is gone through the next process. If the capture image is in poor condition
such as blurring and contains noise, then the corresponding process step certainly cannot achieve its effectiveness. In this research, the image acquisition process is done using a
static camera placement and a static object. There are two cameras used in this research and each of them placed in static. Two possible position of camera are:
1. Top Camera – placed at the top of the object for capturing image in defining x and
y coordinates. Used in Defect Shape Matching DSM, Defect Shape Pointing DSP and Defect Inspection Monitoring DIM.
2. Front camera – placed at the front of object perpendicular to top camera in
defining x and z coordinates. Used in Defect Shape Pointing DSP in providing data of 3D transformation.
Static camera placement is preferred compared to the moving camera because the noise of capturing a moving object is less than the noise of capturing using a moving
camera, however small the movement. The most recommended setup however is to locate the camera directly above or to the side of the object in a static position.
114
4.4.3 Constraint in Workspace Area Size
From the previous discussions, it is known that the image size is important during template creation step, but the workspace size must also be considered carefully during the
training phase. Workspace size refers to the area that the camera can observe; in this case it will be equal to the image shown in the window screen at the GUI. In this research, the
object placed at the centre of the workspace to get the better view of the object. The perfectly place object will result to the better recognition rate while the object placed
slightly out the workspace are will result to poor recognition rate because of the missing characteristic of the object. Therefore care must be taken to ensure that the feature size is
big enough while sufficient workspace size is still able to allow a deviation that might occurs. Figure 4.19 shows an illustration of good captured image within the workspace and
Figure 4.20 shows an image that is partially absent from the workspace.
Figure 4.19: Good Placement of Required Features in Workspace
115
4.4.4 Constraint in Katana Arm Robot