نوع مقاله : مقاله پژوهشی
نویسندگان
1 استادیار، دپارتمان برق و کامپیوتر،دانشکده شریعتی، دانشگاه فنی حرفه‌ای، استان تهران، ایران
2 مهندس تکنولوژی الکترونیک، دپارتمان مهندسی برق و کامپیوتر، دانشکده شریعتی، دانشگاه فنی حرفه‌ای استان تهران، ایران
3 دانشجوی دکتری دانشگاه امیر کبیر
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
The goal of this investigation is weighting and selecting chickens in a determined weight by using machine vision. The images were taken by Raspberrypi cameras and then processed using the Raspberrypi 3 board, and finally the weight was estimated. The marking system was also used to mark chickens in the ideal weight range. For this purpose, 300 Rass chickens were examined in 3 poultry farms by a portable system and from about 500 pictures, 4 features were extracted: length, width, area and circumference. Chickens weight calculated with mixing features (length, width, area) that had been gained. Average percentage error of each one was also measured separately. Average percentage error in this system was 6% (minimum error percentage 2% and maximum error percentage 12%) that indicates the ability of image processing to determine the weight of the chicken. For check system accuracy, one place similar to poultry farms has designed with the same light, temperature, food and water for chickens, and they had access freely, they have been checked for 4 days, then marked chickens were putted up on a digital scale. Finally, the results showed the complete accuracy of the system and the accuracy of the image processing and weighting process
کلیدواژه‌ها [English]
)
