| Summary: | 碩士 === 國立臺北科技大學 === 電機工程系研究所 === 105 === Long-term consumption of un-fresh eggs may cause some health concerns.
Traditionally, people could use sense organs (eyes, ears, nose, and skin) or equipment
to get the freshness of eggs. However, the sense organs were affected by personal
subjective experience. Also, the egg quality measurement equipment is too large for
bringing in daily life. In view of this, we proposed a small-size optical non-destructive
egg quality measurement device. Spectrum processing procedures in this thesis is
divided into 4 stages. In the first stage, we discussed the result of using multiplicative
scatter correction and standard normal variate to reduce the light scattering effect. In
the second stage, we processed the spectrums by the 1st/2nd derivative or fast Fourier
transform to make the features more easier be extracted and be identified. In the third
stage, we discussed the extracted features by principal components analysis and linear
discriminant analysis. In the fourth stage, we used K-nearest neighbors (KNN) to
classify feature and to determine the egg quality. Our experimental results show that
the highest correct rate of KNN can be up to 92.09%
|
|---|
