kyoto university 京都大学 effectiveness of pl filtering image with halogen lamp for agricultural...
TRANSCRIPT
KYOTO UNIVERSITY
京都大学
Effectiveness of PL Filtering Image with Halogen Lamp for Agricultural Products
KYOTO UNIVERSITY
京都大学
Problems on image acquisition
Halation Surrounding walls reflection
KYOTO UNIVERSITY
京都大学
Cuticular layer of apple fruit
Cuticular layer
Epidermis
ParenchymaHalation
KYOTO UNIVERSITY
京都大学
Objectives
• To investigate relation between cuticular layer and fruit surface gloss
• To investigate effectiveness of PL filtering image compared with conventional dome and diffuser-used images
KYOTO UNIVERSITY
京都大学
Dome or diffuser
Diffuser
HalationWall reflection
Light
Camera
Reflection plate
Indirect lighting devices (Dome and Diffuser)
KYOTO UNIVERSITY
京都大学
Direct Lighting Device with PL filter
PL filter Heat absorption filters
Fan(more than 1000nm)
(800 ~ 1000nm)
Halogen lamp
PL filtering
Light
Camera
PL filter
True color
PL filter
KYOTO UNIVERSITY
京都大学
Comparison of lamps
Price Low Low Mid High High
Halogen Incandescent Fluorescent LED HID
Brightness High Mid Low Low High
Color rendering High High Low Mid Mid
Lumen High High Mid High LowMaintenance
Life Mid Short Mid Long Long
KYOTO UNIVERSITY
京都大学
Direct lighting device with LEDs
120 white LEDs
Easy control: Quick Response for input signal
KYOTO UNIVERSITY
京都大学
Lighting Device ArrangementArrangement 1(Dome walls and 4 DLs)
Arrangement 2 and 3(Diffusers or PL filters and 4 DLs)
KYOTO UNIVERSITY
京都大学
Materials
• category 1 (glossy surface fruits: apples, tomato, eggplant, pepper, lemon)
• category 2 (short hair fruits or powdery fruits:
kiwi, peach, pear, banana)
• category 3 (root vegetables: potato, Daikon radish)
KYOTO UNIVERSITY
京都大学
Category 1 (Smooth and thick cuticle)
Walls of dome
Dome Diffuser PL filter
Apple
Eggplant
KYOTO UNIVERSITY
京都大学
Category 2 (Rough cuticle)
Dome Diffuser PL filterPeach
Pear
KYOTO UNIVERSITY
京都大学
Category 3 (No cuticle)
Dome Diffuser PL filter
Daikon radish
Potato
KYOTO UNIVERSITY
京都大学
Relation between degree of gloss and cuticular layer thickness
0
10
20
30
40
50
60
70
80
90
100
0 20 40 60 80 100 120
Thickness of cuticular layer (pixel
Deg
ree
of g
loss
potatoradish
banana
lemon
eggplant
pepper
apples
pearpeach
tomato
Category 1
Category 2
Category 3
(100 pixels is almost 25 m.)
Degree of gloss: SD within twice area of halation
KYOTO UNIVERSITY
京都大学
Images on different exposure conditions
Dome Diffuser PL filter
Exp.1(dark)
Exp.2(mid)
Exp.3(bright)
KYOTO UNIVERSITY
京都大学
HSI on different exposure conditions
Dome Diffuser PL filter
Exp.1(dark)
Exp.2(mid)
Exp.3(bright)
KYOTO UNIVERSITY
京都大学
Hue
Exp.1(dark)
Exp.2(mid)
Exp.3(bright)
Dome Diffuser DL with PL filter
KYOTO UNIVERSITY
京都大学
Hue on different exposure conditions
(Average hue value change except too bright and too dark pixels)
300
310
320
330
340
350
360
370
Exp.1(dark) Exp.2(midium) Exp.(bright)
DiffuserDomePL
KYOTO UNIVERSITY
京都大学
Saturation
Exp.1(dark)
Exp.2(mid)
Exp.3(bright)
Dome Diffuser DL with PL filter
KYOTO UNIVERSITY
京都大学
Saturation on different exposure conditions
020406080
100120140160180200
Exp.1(dark) Exp.2(midium) Exp.(bright)
DiffuserDomePL
(Average hue value change except too bright and too dark pixels)
KYOTO UNIVERSITY
京都大学
Conclusion
• Degree of gloss (halation) on fruit had proportional relation with thickness of cuticular layer.
• PL filtering image had higher color reproduction compared with diffuser or dome.
• Dome made larger low saturation area.