Study on detection of crack on the quail egg shell using image processing and neural networks

Abstract

This paper proposes a new method for detecting cracks on the quail egg shell using image processing and  neural networks. Histogram of many images of candling quail egg are used as training data for neural networks to check the accuracy of the proposed method. First, the white light, yellow and red of LED Luxeonwas are used to test and select the appropriate light for the illumination of quail eggs. In particular, the result are  better for the yellow light. Cracks appear clearly, bias histogram of images of candling quail egg between egg cracks and intact eggs are  larger than the red and the white light. Initial results can be achieved with an accuracy of 85,1% for the eggs without cracking, 87,98% for the case of cracks` and the average accuracy  is  86,54%.