Vision Measurement Vision measuring machines mainly provide the following processing capabilities. • Edge detection Quick Guide to Precision Measuring Instruments
Vision Measuring Machines
Dimensional Measurement An image consists of pixels. If the number of pixels in a section to be measured is counted and is multiplied by the size of a pixel, then the section can be converted to a numeric value in length. For example, assume that the total number of pixels in the horizontal size of a square workpiece is 300 pixels as shown in the figure below. If a pixel size is 10 μm under imaging magnification, the total length of the workpiece is given by 10 μm×300 pixels=3000 μm=3 mm.
Detecting/measuring edges in the XY plane
10 µm
300 pixels
• Auto focusing
Focusing and Z-axis measurement
Edge Detection How to actually detect a workpiece edge in an image is described using the following monochrome picture as an example. Edge detection is performed within a given domain. A symbol which visually defines this domain is referred to as a tool. Multiple tools are provided to suit various workpiece geometries or measurement data.
Display screen
244 241 220 193 97 76 67 52 53 53 243 242 220 195 94 73 66 54 53 55 244 246 220 195 94 75 64 56 51 50 The edge detection system scans within the tool area as shown in the figure at left and detects the boundary between light and shade. Example of numeric values assigned to pixels on the tool
Tool
• Pattern recognition
PC
CCD Camera lens
Alignment, positioning, and inspecting a feature Video signal
High-speed A/D converter
Frame grabber
Amplifier
Image Storage
255 127 0
640 pixels
(1) Scan start position (2) Edge detection position (3) Scan end position
Tool position
(2) (3)
(1)
High-resolution Measurement
An image is comprised of a regular array of pixels. This is just like a picture on fine plotting paper with each square solid-filled differently.
When enlarged...
Gray Scale A PC stores an image after internally converting it to numeric values. A numeric value is assigned to each pixel of an image. Image quality varies depending on how many levels of gray scale are defined by the numeric values. The PC provides two types of gray scale: two-level and multi-level. The pixels in an image are usually displayed as 256-level gray scale.
Tool position
Tool position
A position the system recognizes as an edge may be in error by up to one pixel width using normal image processing. This will prevent the execution of high-resolution measurement.
To increase the accuracy in edge detection, sub-pixel image processing is used. An edge is detected by determining an interpolation curve from adjacent pixel data as shown below. As a result, it allows measurement with a resolution better than 1 pixel.
2-level gray scale
Multi-level gray scale
255
White
White
1
Gray
Gray
127
M
0
Black
Black
0
Tool position
Tool position
Each pixel is displayed as one of 256 levels between black and white. This allows high- fidelity images to be displayed.
Pixels in an image brighter than a given level are displayed as white and all other pixels are displayed as black.
The image signal profile approaches an analog waveform like this.
Image signal without sub-pixel processing
Image signal with sub-pixel processing
M-15
Powered by FlippingBook