This paper proposed a method to detect the quality of salted duck eggs based on image processing and artificial neural network. Fourier transform was employed after pre-processing for the images of eggs. It was found that the texture of phase spectrum images of four kinds eggs ("non-well pickled", "well-pickled", "over-pickled" and "spoilt" eggs), varied from clear and ordered to vague and disordered, while the corresponding entropies of images gradually decreased. Through the brightness statistics of images with wavelet reconstruction, it was found that in terms of brightness, the "well-pickled eggs" ranked the first, and the "spoilt eggs" being the last. Color, texture, and statistical moment parameters were selected as image parameters, and BP neural network was created with three-layer feedforward structure (10-18-4). Then the network was tested and the correct ratio of the model amounted to about 92%, certifying the feasibility of this method. © 2011 Springer-Verlag.
CITATION STYLE
Chen, Z., Wang, Q., & Zhu, T. (2011). Non-destructive line detection of salted egg based on image processing and BP neural network. In Communications in Computer and Information Science (Vol. 201 CCIS, pp. 481–488). https://doi.org/10.1007/978-3-642-22418-8_68
Mendeley helps you to discover research relevant for your work.