{"id":{"repo_id":"slu","oai_identifier":"oai:pub.epsilon.slu.se:7"},"canonical_url":"https://search.dev.ndltd.org/etd/slu/oai:pub.epsilon.slu.se:7","repository":{"repo_id":"slu","name":"Swedish University of Agricultural Sciences","base_url":"https://pub.epsilon.slu.se/cgi/oai2"},"display":{"title":"Automatic sorting of sawlogs by grade","abstract":"Optical log scanners provide a possibility for sawmills to pre-sort logs automatically by grade. Logs were classified by different grades and sawn timber properties using external log geometry variables such as taper, surface unevenness, sweep and out-of-roundness. Raw data from commercial shadow scanners and laser point 3D-scanners were used. The analyses were based on data from three sawing studies on Norway spruce, a total of 1095 logs from 22 different stands, and one sawing study on Scots pine. Logistic regression was used as the classification method, and model accuracy was assessed using the areas under receiver operating characteristic (ROC) curves. Models for sorting criteria based on knot size, knot type and grain distortion, like visual stress grades, showed a better predicting performance than commodity grade and MSR (stiffness) models. The results generally improved when variables generated from 3D-scanner data were used. Similar external geometry variables proved useful in the different studies, even if the parameter estimates differed between stands and regions. A recommendation is that mills should verify and develop classification models for their own log supply and sorting criteria. The opportunity for sawmills to use the methods to improve revenue was shown in a glulam case study, where for example high stress grades were requested. The sorting accuracy, and revenue, increased when non-geometry variables such as density and grain angle were adde","abstract_html":"Optical log scanners provide a possibility for sawmills to pre-sort logs automatically by grade. Logs were classified by different grades and sawn timber properties using external log geometry variables such as taper, surface unevenness, sweep and out-of-roundness. Raw data from commercial shadow scanners and laser point 3D-scanners were used. The analyses were based on data from three sawing studies on Norway spruce, a total of 1095 logs from 22 different stands, and one sawing study on Scots pine. Logistic regression was used as the classification method, and model accuracy was assessed using the areas under receiver operating characteristic (ROC) curves. Models for sorting criteria based on knot size, knot type and grain distortion, like visual stress grades, showed a better predicting performance than commodity grade and MSR (stiffness) models. The results generally improved when variables generated from 3D-scanner data were used. Similar external geometry variables proved useful in the different studies, even if the parameter estimates differed between stands and regions. A recommendation is that mills should verify and develop classification models for their own log supply and sorting criteria. The opportunity for sawmills to use the methods to improve revenue was shown in a glulam case study, where for example high stress grades were requested. The sorting accuracy, and revenue, increased when non-geometry variables such as density and grain angle were adde","abstract_has_math":false,"creators":["Armas Jäppinen"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2000,"date_issued":"2000-01","date_published":"2000-01","updated_at":"2026-07-24T04:35:38Z","subjects":["picea abies sawlogs grading sorting equipment computer applications wood properties wood defects","classification","geometry","grade","logistic regression","log","Picea abies","saw scanner","sorting."],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://pub.epsilon.slu.se/7/","outbound_label":"Repository record","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2000-01"]},{"key":"dc:type","label":"Dc Type","values":["Doctoral thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["picea abies sawlogs grading sorting equipment computer applications wood properties wood defects","classification","geometry","grade","logistic regression","log","Picea abies","saw scanner","sorting."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://pub.epsilon.slu.se/7/","https://pub.epsilon.slu.se/7/1/91-576-5873-0.fulltext.pdf"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.other","label":"Dc Description Other","values":["Optical log scanners provide a possibility for sawmills to pre-sort logs automatically by grade. Logs were classified by different grades and sawn timber properties using external log geometry variables such as taper, surface unevenness, sweep and out-of-roundness. Raw data from commercial shadow scanners and laser point 3D-scanners were used. The analyses were based on data from three sawing studies on Norway spruce, a total of 1095 logs from 22 different stands, and one sawing study on Scots pine. Logistic regression was used as the classification method, and model accuracy was assessed using the areas under receiver operating characteristic (ROC) curves. Models for sorting criteria based on knot size, knot type and grain distortion, like visual stress grades, showed a better predicting performance than commodity grade and MSR (stiffness) models. The results generally improved when variables generated from 3D-scanner data were used. Similar external geometry variables proved useful in the different studies, even if the parameter estimates differed between stands and regions. A recommendation is that mills should verify and develop classification models for their own log supply and sorting criteria. The opportunity for sawmills to use the methods to improve revenue was shown in a glulam case study, where for example high stress grades were requested. The sorting accuracy, and revenue, increased when non-geometry variables such as density and grain angle were adde"]},{"key":"dc:title","label":"Title","values":["Automatic sorting of sawlogs by grade","Acta Universitatis Agriculturae Sueciae. Silvestria"]}]}],"canonical_facts":{"dc:date.issued":["2000-01"],"dc:description.other":["Optical log scanners provide a possibility for sawmills to pre-sort logs automatically by grade. Logs were classified by different grades and sawn timber properties using external log geometry variables such as taper, surface unevenness, sweep and out-of-roundness. Raw data from commercial shadow scanners and laser point 3D-scanners were used. The analyses were based on data from three sawing studies on Norway spruce, a total of 1095 logs from 22 different stands, and one sawing study on Scots pine. Logistic regression was used as the classification method, and model accuracy was assessed using the areas under receiver operating characteristic (ROC) curves. Models for sorting criteria based on knot size, knot type and grain distortion, like visual stress grades, showed a better predicting performance than commodity grade and MSR (stiffness) models. The results generally improved when variables generated from 3D-scanner data were used. Similar external geometry variables proved useful in the different studies, even if the parameter estimates differed between stands and regions. A recommendation is that mills should verify and develop classification models for their own log supply and sorting criteria. The opportunity for sawmills to use the methods to improve revenue was shown in a glulam case study, where for example high stress grades were requested. The sorting accuracy, and revenue, increased when non-geometry variables such as density and grain angle were adde"],"dc:identifier.uri":["https://pub.epsilon.slu.se/7/","https://pub.epsilon.slu.se/7/1/91-576-5873-0.fulltext.pdf"],"dc:subject":["picea abies sawlogs grading sorting equipment computer applications wood properties wood defects","classification","geometry","grade","logistic regression","log","Picea abies","saw scanner","sorting."],"dc:title":["Automatic sorting of sawlogs by grade","Acta Universitatis Agriculturae Sueciae. Silvestria"],"dc:type":["Doctoral thesis"]},"updated_at":"2026-07-24T04:35:38Z"}