Merjenje volumna z globinsko kamero Intel RealSense D435

Lukančič, Žiga (author), Mihelič, Matevž (mentor)

Pri ugotavljanju prostornine sekancev na kupih ali prikolicah se srečujemo z vprašanjem točnosti ocene njihove prostornine. Trenutno uveljavljene metode se osredotočajo na metode ekspertnih ocen, kjer izkušeni merilci ocenjujejo količino glede na svoje predhodne izkušnje. Metodo smo skušali izboljšati z uporabo sodobne globinske kamere Intel RealSense D435, ki za merjenje globine uporablja LIDAR stereo senzorja. Sama kamera deluje kot senzor, zato smo pripravili še aplikacijo, ki omogoča izvedbo meritev v realnem času. V aplikacijo lahko vnesemo podatke o velikosti prikolice, program pa glede na podatke iz senzorja in slike prazne prikolice ter velikosti modela kupa izračuna velikost kupa. Tako pripravljeno aplikacijo smo testirali na modelu kupa znane prostornine, pri čemer smo kontrolirali tudi osvetljenost. Volumen modela kupa znaša 0,0976 m3. Pri prvih meritvah smo izračunali volumen 0,0938 m3 ± 0,0051 m3, pri drugih meritvah 0,1088 m3 ± 0,0086 m3 in pri dodatnih meritvah 0,1012 m3 ± 0,0101 m3. Izmera z globinsko kamero je imela natančnost od 3,68 % do 11,45 %. To je manj od pričakovane natančnosti, ki jo za kamero navaja njen proizvajalec (<2%).

• Language: Slovenian
• Keywords: lesni sekanci, LIDAR
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: BF - Biotechnical Faculty
• Publisher: [Ž. Lukančič]
• Year: 2024
• PID: 20.500.12556/RUL-159217
• UDC: 630*52(043.2)=163.6
• COBISS.SI-ID: 200672259
• Publication date in RUL: 04.07.2024

Secondary language

• Language: English
• Title: Volume measurment using depth camera Intel RealSense D435

Abstract

When measuring the amount of wood chips, we face the issue of accurately determining the volume of the piles on the ground or the contents of the trailer. Currently established methods focus on expert estimation methods, where experienced measurers estimate the quantity based on their previous experience. We tried to improve the method by using a modern depth camera Intel RealSense D435, which uses LIDAR stereo sensors to measure depth. The camera itself works as sensor, so we have prepared an application below that enables measurements to be taken in real time. We can enter information about the size of the trailer into the application, and the program calculates the size of the pile based on the data from the sensor and images of the empty trailer and the size of the model pile. We also tested the application prepared in this way on a model pile of known volume. Volume of model pile is 0,0976 m3. For the first measurements we calculated the volume of 0.0938 m3 ± 0.0051 m3, in second measurements 0.1088 m3 ± 0.0086 m3 and in additional measurements 0.1012 m3 ± 0.0101 m3. The measurement with a depth camera had a lower accuracy as stated by the manufacturer (< 2 %).

• Keywords: wood chips, LIDAR