摘要
针对颜色色空间转换非线性的复杂关系,在获取标准阈值颜色色度值,进行归一化处理后,利用学习矢量量化(LVQ)神经网络和概率神经网络(PNN)进行尿样颜色识别。比对结果表明:(1)利用神经网络进行分类识别时实验数据的归一化处理是完全必要的。(2)与颜色色差评价方法进行了比对,该方法可行而有效。
In light of nonlinear complex relationship in the color space transformation, after the standard threshold color chromaticity values are normalized, Learning Vector Quantization (LVQ) neural network and Probabilistic Neural Network (PNN) are used to identify the urine color. The comparison results show that: (1) When neural network is used to classification and recognition, the normalization of experiment data is absolutely necessary. (2) Compared with the color difference evaluation method, this method is feasible and effective.
出处
《绥化学院学报》
2013年第8期165-168,共4页
Journal of Suihua University
基金
黑龙江省教育厅科学技术研究项目(12511631)
绥化学院科学技术研究项目(K1002002)