Accuracy, efficiency and usability of photo-optical pile measurement systems in the wood supply chain

Transcription

1 Accuracy, efficiency and usability of photo-optical pile measurement systems in the wood supply chain Martin Opferkuch September 27, 2017 Martina Attinger, Dirk Jaeger

2 Context Several systems for hand-held photo-optical measurement of roadside CTL log piles on the market Tablet or smartphone w/ built-in camera Series of overlapping pictures stitched together Automatic detection of single logs by algorithm Manual post-editing possibility/necessity Output Pile volume Number of logs Diameter distribution 2

3 Objective Can these systems compete with traditional manual measurements regarding Accuracy? Time consumption? Costs? Are there differences between the systems? 3

4 Testing scenario (I) 5 photo-optical measurement systems 4 assortments spruce sawlogs 5 m length larch sawlogs 5 m length spruce industrial wood 3 m length beech industrial wood 3 m length 3 piles per assortment ~ 25 m³, ~ 50 m³, ~ 75 m³ Very accurate piling w/ exact harvester measurement for each pile 3 measurements per pile & system 4

5 5 Systems tested Company Product SCHELLER SYSTEMTECHNIK AFoRS FOVEA UG ifovea GDD-IT GmbH PolterMAX-2D HEIDEGESELLSCHAFT FORST- PRODUKTE UND -GERÄTE GmbH Tradenda IABG WoodCa Hardware, Operation system and additional required material Smartphone or Tablet w/ GPS + logger s tape + measuring set with marker & fixation material Smartphone or Tablet ios or Android w/ GPS + logger s tape + colour spray Windowstablet w/ GPS + logger s tape Smartphone or Tablet (Android OS) w/ GPS + logger s tape + colour spray + measuring staff Windows Tablet /Toughpad w/ GPS + logger s tape Computation Server based Tablet based Tablet based Server based by service provider Tablet based 5

6 Data acquisition 6

7 Volume [m³ sub] Volume [m³ sub] Results: volume (I) Ø % +8% Volume difference larch sawlogs Ø % +3% SL31 SL32 SL33 Larch log piles -4% +17% Ø Harvester Harvestermaß Traditional Stirnflächenverfahren Mill Werkseingangmaß m 3 sub / log Volume difference spruce sawlogs Ø % -2% Ø % +6% Ø % +16% Harvester Harvestermaß Traditional Stirnflächenverfahren Mill Werksmaß SF11 SF12 SF13 Spruce log piles 7

8 Percentage [%] Results: volume (II) spruce larch Volume difference in relation to mill measurement [%]?? larch average SL (gemittelt) SF (gemittelt) spruce average Systems 8

9 percentage [%] Results: number of logs spruce larch n=125 n=212 n=310 n=85 n=140 n= Log count difference in relation to mill measurement [%] larch average SL (gemittelt) SF (gemittelt) spruce average 9

10 average time [min] / m 3 sub AFoRS ifovea Poltermax Tradenda WoodCA Manual AFoRS ifovea Poltermax Tradenda WoodCA Manual Results: time consumption average time consumption per assortment SF Systems SL capturing time post-editing 10

11 Testing scenario (II) Additionally 1 test pile of ~150 m³ spruce sawlogs with specific sections of potential difficulties for log detection Percentage deviation of correct detection from the reference Branches Bark Text Dots Offset Offset Offset deviation Dirt 11 Offset

12 deviation deviation Results: Examples of disturbing factors Bark pieces: percentage deviation of correct detection from the reference PolterMAX (auto) ifovea (auto) PolterMAX (auto) WoodCA (auto) Dirt: percentage deviation of correct detection from the reference 12

13 Conclusion Photo-optical measurement systems can be better than manual systems under good conditions Regarding volume (and diameter distribution) Documentation Good integrateability in WSC through digital data Closing of data gap proofing forwarding performance Large differences in performance between systems Need for careful piling Necessity of post-editing Costs mostly depend on time consumption 13

14 Acknowledgement Forstliches Bildungszentrum für Waldarbeit und Forsttechnik Regionalforstamt Hochstift Martin Opferkuch Chair of Forest Operations Telefon