Development of a temperature measurement system for a broiler flock with thermal imaging
| Autor: | Ya-Cheng Liu, Ying-Jen Haung, Perng-Kwei Lei, Pao-Nan Shen, Jeng-Liang Lin |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
business.industry
General Chemical Engineering 0402 animal and dairy science Broiler Automatic threshold Image processing 04 agricultural and veterinary sciences 040201 dairy & animal science Temperature measurement Industrial and Manufacturing Engineering Gaussian filter symbols.namesake Thermal 040103 agronomy & agriculture symbols 0401 agriculture forestry and fisheries Raised temperature Computer vision Flock Artificial intelligence business Simulation Food Science Mathematics |
| Zdroj: | Engineering in Agriculture, Environment and Food. 9:291-295 |
| ISSN: | 1881-8366 |
| DOI: | 10.1016/j.eaef.2016.03.001 |
| Popis: | Broiler body temperature is an important indicator of both a broiler's response to its environment and its overall health. This study used an infrared thermal camera to capture the thermal images of a broiler flock in a broiler house and then analyzed the temperatures of the broiler flock by image processing. Given the lack of feathers on a broiler's head and feet, the thermal images showed a raised temperature for these parts, making the head an appropriate part for measuring body temperature. The average or Gaussian filter was first used to smooth the images, after which the Otsu automatic threshold algorithm was used to determine the gray-scale thresholds and to segment the broiler's head and feet, followed by segmenting of the overlapping broiler images via the watershed method. Area filtering was then used to identify the broiler head blocks among the segmented blocks. Finally, the labeling methodology was used to calculate the number of broilers in each image. Analysis results show that the proposed image processing procedures can successfully identify the broiler head and analyze the temperatures of the broiler flock by capturing thermal images from a height of 160 cm and a depression angle of 30°, while achieving an identification accuracy of up to 91.3%. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|