AL1_SiplaceX-en.pdf - 第296页

1 - 6 S tudent Guide SIPLACE X 8 Siplace Vision Edition 02/2005 6 8.1.5 Synthetic fiducials In rough se arch step is with reduce d resolution th e gray va lue difference betwe en both po ints of a point pair de termined.…

1 - 5

Student Guide SIPLACE X

Edition 02/2005 8 Siplace Vision

8.1.4 Shape of fiducials for PCB position and placement position recognition

Following 7 (9) shapes of fiducial could be taught as a synthetic fiducial.

Following measuring results of the fiducial issued:

Kind of fiducial (naming in fiducial

list) What to teach / comment

Cross; Double cross; Shape, dimension, stroke thickness and bright or dark picture.

Circle Shape, dimension and bright or dark picture..

Circle ring Shape, dimension, stroke thickness and bright or dark picture.

Rectangle or square (square a special

form of a rectangle)

Shape, dimension and bright or dark picture.The square is a well-known

special case of the rectangle where all sides are adjusted to identical

values.

Rectangular ring or squared ring(squared

ring a special form of a rectangle ring)

Shape, dimension, stroke thickness and bright or dark picture.The

squared ring is a well-known special case of the rectangular ring where

all sides are adjusted to identical values.

'Diamond' - a 45° turned square Shape, dimension and bright or dark picture.

'Pattern' other fiducial shapes 2D (trian-

gle, circle segment, conductive pattern or

similar structures)

Pattern recognition algorithm, the fiducial is as a - '2 D-fiducial' - in the

fiducial list named.

At fiducial teaching: At fiducial centering:

=> X-position of the structure to camera center => X-position of the structure to camera center

=> Y-position of the structure to camera center => Y-position of the structure to camera center

=> Angle deviation of the structure to PCB-camera

coordinate system

1 - 6

Student Guide SIPLACE X

8 Siplace Vision Edition 02/2005

8.1.5 Synthetic fiducials

In rough search step is with reduced resolution the gray value difference between both points of

a point pair determined. The shape position of the point pair calculates the algorithm from the geo-

metric model.

From the gray value difference the real position is determined.

In fine search step is along the scanning lines the largest gray value difference determined. From

this positions is the regression line calculated their position define the fiducial center position. The

positions of the scanning lines are defined from the point pair positions of the rough search. Par-

allel to the position measurement is a quality value determined. This quality value show how ac-

curate fit the theoretical model at the determined fiducial position.

This principal could be used for bad mark recognition too. The negative bad mark is recognized if

the quality level is lower than a defined limit.

Camera picture of the

fiducial

Rough search with reduced

resolution:

Position of the

scanning lines (yellow) at rec-

ognized pairs.The blue point

has a higher gray value

(brighter) than the yellow one.

Fine search: Position of the

scanning lines (yellow) at

recognized pairs.

Green crosses: found posi-

tions of the largest gray gra-

dient along the Scanning line

Green straight line: regres-

sion line

1 - 7

Student Guide SIPLACE X

Edition 02/2005 8 Siplace Vision

8.1.6 Pattern as fiducials

Learn patterns with an automatic algorithm for 2D fiducial:

=> first is the search area the so-called Region Of Interest ROI defined.

=> the reference point of the pattern have to be calculated (recommended) or manual selected

(more imprecise with this manual definition). (This is the point where the placement refers to).

=> the pattern is calculated to determine the reference point.

=> test pattern on reliability using a few different individual fiducials (stability of measuring results).

The precision of the Reference point defines the placement accuracy later. Best is to determine

the reference point automatically and check it.

8.1.7 Bad mark recognition

Bad mark recognition could be programmed on following different ways.

8.1.8 Synthetic bad mark

There are 9 different synthetic shapes offered for synthetic bad mark. Similar teaching fiducials:

the shape, the dimension and the optical impression - bright or dark - of the 'positive bad mark'

have to be taught. Is the shape recognized the positive bad mark for placement is recognized. Is

the shape in search area not recognized is a negative bad mark recognized (no time waste at

missing structure) and no placement is done respective as a global bad mark the local bad mark

of the single circuits are measured. (See 4.1.1)

This -left- picture shows the fine search of the

synthetic Bad Mark teachings with result.

This -right- picture is the result of the negative

marked bad mark.

Camera picture of trian-

gle fiducial

Defined reference point:The

reference point of a fiducial

defines the placement positions.

Therefore this has to be as pre-

cise as possible because the

basic coordinates of the PCB-

position are set here. -Here the

position with the largest bright-

ness difference was selected -.

Learned pattern:

Referring to the reference

point learn the system the

shape of pattern and save

this outline as 'Teach pic-

ture'.