Smart Gui - Algorithm Training Material_rev1.pdf - 第140页

Copyright © ViT rox All Rights Reserved . Lead Ringer - Parameter Description 140 1. Compare T o ● This option enable user to setup the comparison option ● Comparison “between leads” or “against an absolute threshold” ● …

Lead Ringer - Default Pretest

Page 1: Pad Too Dark

• Inspect Pad Region, If lifted lead, end of the pad will look

darker

Page 2: Toe Fillet Curvature Too Low

• Inspect Lead Tip Region, If the lead is lifted, the solder shape

will not be the same

Page 3: Toe Fillet Shape Differs

• Inspect Lead Tip Region. The solder shape varies if the lead

is lifted

Page 4: Shoulder Bend Differs

• Inspect Shoulder Region. If the lead is lifted (heavily) or bent,

• The lead will be visibly deformed: the shoulder will appear

brighter in the current region of interest.)

139

Lead Ringer - Parameter Description

140

1. Compare To

● This option enable user to setup the comparison option

● Comparison “between leads” or “against an absolute threshold”

● 3 options as follow:

➔ Other Leads: Each lead is compared to all the other leads of the same row. The lead that

get a score 0 is recognized as the "base" of the row. We look at the difference between

leads.

➔ Other Leads (Invert Check): The invert check will compare leads that are failing against a

good template build from previous inspected components and determine if it is similar to it or

not (the threshold used to do the comparison and determine if there is a switch needed is

the threshold from the concerned page). If no leads are failing, we do not perform the invert

check.

➔ Threshold: Each lead is compared to the absolute threshold given by the user (separately

from the other leads of the row).

Lead Ringer - Parameter Description

141

2. Extract

● The Extract option available depends on the “Compare To” option selected.

● If “Compare To” is set to “Other Leads” & Other Leads (Invert Check), There are 3

Extract Options as follow:

➔ Mean Intensity Difference: we measure the mean intensity within the box and calculate the

difference with mean intensity of the "base" lead of the row (median lead). If the difference is

bigger/smaller of the threshold (depending on the threshold sense (high bad/low bad)), than

the lead will be called defective.

➔ Appearance Difference: we compare each pixel contained in the current box with the

corresponding pixels in the box of the "base" lead and calculate its difference. Then each

difference is summed up for all the pixel to produce an unique score.

➔ Curvature Difference: we calculate the curvature feature (distance between red channel

(Angle 1) and green channel (Angle 2) centre of gravity) in the current box, and then

measure the difference between the resulting value and the "base" lead of the row