weight estimation system of individual broiler chickens using digital image processing and multi-regression analysis

the purpose of this study was to identify daily changes in body weight of broiler chickens using image processing and multi regression analysis. therefore thirty 1 day old broiler chickens were reared under standard rearing condition and after acquiring images they were weighted individually. from 2490 digital images six features (perimeter area convex major axis minor axis eccentricity) were extracted. linear equations between body weight and these features indicated that r2 values for these features (except for eccentricity) for the individual birds were higher than 0.9. furthermore stepwise selection process was utilized to develop multi regression model and to remove non significant factors from the regression equation. according to the developed equation area perimeter area convex major axis minor axis interaction between area and major axis and convex area and perimeter major and minor axis were capable of predicting weight with r2= 0.945 in the confidence level of %5. this shows that the digital image processing and multi regression analysis could predict weight of life chickens promisingly.