Study On Growth Curve Fitting Method Of Chicken Based On Depth Image

The broiler growth model was obtained by using the actual weight and the three typical growth curves of poultry.The traditional broiler weight weighing was manually weighed to the electronic scale and was in direct contact with the broiler,which was not conducive to its welfare.At home and abroad using machine vision technology in large animal body weight estimates on the results,but the poultry small posture changes quickly,not easy to use traditional machine vision technology to estimate.Through the combination of Kinect 3D camera and wireless platform,the data of the obtained broiler were used to establish the broiler weight model and then the growth curve was fitted to obtain the optimal growth curve of the experimental group.And then for the current stage of complex weighing chickens,welfare and other issues,more convenient and clear the growth and development of broiler law.This paper first set up a set of broiler weight and depth image acquisition system,the use of electronic scale trigger camera to shoot,when the chickens appear in the experimental fence will be based on the relevant settings automatically completed the broiler weight data and depth of the image acquisition,reduce the experimental process The loss of manpower.Using Matlab software to shoot the depth of the picture preprocessing and feature extraction.In this paper,the image preprocessing part is transformed into the binary image which is conveniently processed by image cutting,median filtering,threshold segmentation,and depth processing.Then,the image preprocessing part is extracted from the one-dimensional,two-dimensional and three-dimensional The feature is used to construct the model later.According to the characteristics of pre-extraction and related weight,two kinds of broiler body weight grading methods were constructed by using SVM classifier and RBF neural network.The grading results were compared and analyzed.The classification of broiler weight based on depth images was completed,The results showed that SVM had better effect on the weight of broiler.Breaking the traditional broiler weight and weighing the status of broiler contact,and then designed a 50-day broiler weight modeling experiment,the use of BP neural network to construct the broiler weight estimation model,and compared the characteristics of the model to establish the contribution rate,the results show The estimated height of broiler body weight was higher by BP neural network.Using the modeling experiment to construct the broiler weight estimation model,a 50-day fitting experiment was designed.The chicken broth obtained by fitting the experiment was input into the broiler weight model obtained by modeling experiment,and the body weight of the broiler was estimated.SPSS software was used to fit the estimated weekly weight trend graph and three typical nonlinear growth curve models of poultry,and the growth curve model of the green pheasant was obtained with the optimal fitting degree.Through this method,the optimal growth model of broiler was obtained,which reduced the loss of manpower and further improved the welfare of broiler.