Hatching Eggs Online Monitoring System Based On Machine Vision

The hatching rate is affected not only by the hatching egg quality whethergood or bad, but also by the quality of newborn chicks, the future health of thepoultry and the power of life and performance. The activity detection of embryo isan important technical aspect in the egg hatching process. In view of thelabor-intensive, low efficiency and poor accuracy in manual detection. Anautomatic and practical detection system based on machine vision system isdeveloped instead of manual inspection of hatching egg for improving detectingaccuracy and efficiency.This dissertation firstly introduces the research background and developmentof egg hatching activity and machine vision. Then, the egg hatching activitydetection hardware system is built, and this dissertation analyses the basic concepts,framework and key technique of the image enhancement, image segmentation andregion labeling, and then the color model is selected. Finally, this dissertationpresents the hatching detection method. The simulated annealing particle swarmalgorithm and BP neural network are described in a detail, and then the keytechnique of the neural network is further analyzed and the optimal algorithm isproposed to improve the parameters of BP neural network. This dissertation studiedfrom several aspects, the main work is as follows:1． The machine vision hardware system is build for identifying exterior qualityand fertility of hatching egg. The light source and background color are foundout through a lot of experiments.2． The eggs screening methods based on machine vision technique has beensystematically studied. Color model and image pre-processing methods lay thefoundation for further neural network training.3． An improved GA-SA is put forward, which is used to optimize topologystructure of BP neural network for detecting fertility of hatching eggautomatically. Detection method of fertility of hatching egg during the entirehatching period is developed systematically. Hue、brightness and saturationfrequency values of the hatching egg image is selected as the input of neuralnetwork. Fertility of hatching egg is identified by the improved GA-SA neuralnetwork. The neural network system for fertility of hatching eggs detection has a high accuracy and generalization ability.4． A software package for all functions is integrated finally which lay afoundation for further study.