[PPT] - FamilySearch Scanning (Scanstone) An Automated Exposure Method For PowerPoint Presentation

SLIDE 1

FamilySearch Scanning

(Scanstone)

An Automated Exposure Method For Scanning Microfilm

Heath Nielson nielsonhe@ldschurch.org

SLIDE 2

Digitizing Microfilm

Scan as efficiently as possible

– At 30 minutes a roll, it would require a single

scanner 142 years of constant scanning to scan 2.5 million rolls

Maintain consistent (good) quality

– Adjust lamp level

SLIDE 3

Changing light level

Lamp level: 80 180 200 210 220 230 240 250 90 100 110 120 130 140 190 150 160 170

SLIDE 4

Goal

To maintain efficiency and quality, we

need to automatically identify the optimal lamp setting for a given film

Optimal lamp level is dependent upon:

– Film density – Focal length – Light output

SLIDE 5

Film Density

Amount of incident light transmitted through the film

is measured in terms of optical density

∞ ≤ ≤ D ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ =

t i

I I D log

) ( i I ) ( t I

28 . 07 . − ≈

min

D 80 . 1 80 . − ≈

max

D

(85%-52%) (15%-1%)

D

Film Contrast is dependent upon range between and

max

D

min

D

SLIDE 6

Density vs. Lamp Level

Histogram View

50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity 50 100 150 200 250 300 8 9 5 1 1 1 2 5 1 4 1 5 5 1 7 1 8 5 2 2 1 5 2 3 2 4 5 Lamp Intensity

16 . ≈

min

D

40 . 1 ≈

max

D

SLIDE 7

Density Test

Created test film comprised of 17 sections

– The density of each section was constant – The sections represented a range of densities

from 0.03 to 1.50

Film was scanned multiple times across a

range of different lamp levels

The mean intensity for each density was

calculated

SLIDE 8

Density Roll

SLIDE 9

Result of Test Roll

200DPI-16x

0.0 50.0 100.0 150.0 200.0 250.0 300.0 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Lamp Intensity 0.03 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.30 1.40 1.50

SLIDE 10

Focal length

Camera Lens

Camera and lens move

– Affects the amount of light

received by the camera

– Position is determined by

the selected DPI and reduction ratio of the film prior to scanning

SLIDE 11

Focal Length vs. Lamp Level

200DPI-10x

0.0 50.0 100.0 150.0 200.0 250.0 300.0 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Lamp Intensity 0.03 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.30 1.40 1.50

200DPI-12x

0.0 50.0 100.0 150.0 200.0 250.0 300.0 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Lamp Intensity 0.03 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.30 1.40 1.50

200DPI-16x

0.0 50.0 100.0 150.0 200.0 250.0 300.0 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Lamp Intensity 0.03 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.30 1.40 1.50

200DPI-24x

0.0 50.0 100.0 150.0 200.0 250.0 300.0 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Lamp Intensity 0.03 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.30 1.40 1.50

200DPI-42x

0.0 50.0 100.0 150.0 200.0 250.0 300.0 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Lamp Intensity 0.03 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.30 1.40 1.50

SLIDE 12

Scanner Light Output

Light output is

inconsistent across scanners

Mechanical differences

– Presence of diffuser – Out of calibration – Age of lamp

Scanners Compared - Lumens

0.00 200.00 400.00 600.00 800.00 1000.00 1200.00 1400.00 64 128 192 255 Lamp Level Lumens S01 S02 S03 S04 S05 S06 S07 S08 S09 S10

SLIDE 13

Developing an Exposure Strategy

Input

– Unknown film density – Variable focal length – Inconsistent light output across scanners

Output

– Inter-frame background and handwriting with an

intensity ~0

– Frame background with an intensity ~200 or greater

How should the lamp be adjusted?

SLIDE 14

Histogram Stretching

Set the lamp to an “average” value for the

selected focal length

Scan the film Find the mode of the black and white

distributions

Create a stretching function to remap the

scanned pixel data

) , (

w b x

x

) (

b

x x a y − =

b w

x x a − = 200

SLIDE 15

Example

177 =

b

x 857 . 2 177 247 200 = − = a

) 177 ( 857 . 2 ) ( − = x x f

247 =

w

x

SLIDE 16

Invariant to minor lamp changes

SLIDE 17

Potential problems

Still some dependence upon lamp level

– Extreme values cause camera saturation

Assumes a single black or white

distribution

– What value produces acceptable results for a

multi-modal distribution?

Fewer gray levels by stretching the image

data post-scan

SLIDE 18

Manual Audit

A random sample of 469 films were

selected from the 2.5 million rolls

Each film was scanned and audited for

quality

If quality was deemed unacceptable, the

contrast could be overridden

SLIDE 19

Results

185 (39%) films were manually adjusted

– As the test proceeded, the number of films

adjusted decreased

– In 155 films (84%), the contrast was

increased

Average number of gray levels

– 187 before audit – 169 after audit

SLIDE 20

Future Directions

Better metrics for determining a ‘good’ quality

image

Localized histogram stretching

– Per region – Per frame

12-bit scanning

– Decreases further, dependence upon lamp – Increases likelihood of capturing full dynamic range

between and

min

D

max

D

SLIDE 21

Conclusion

We have shown an exposure algorithm

which produces consistent output in the presence of differing amounts of light and is automated, maintaining the efficiency necessary to scan large numbers of film