基于形状先验水平集的高鲁棒性中医舌图像分割方法

中医舌图像分割一直面临着由于采集设备更新换代等因素带来的分割方法鲁棒性下降的问题.提出一种基于形状先验主动轮廓模型的中医舌图像分割方法.首先,采用统计方法获得大致舌体图像;然后,采用形状先验水平集方法得到基本完整的舌体;最后,采用Snake模型方法得到最终准确分割结果.实验结果表明:该方法对于不同环境下得到的舌图像分割都具有良好的分割效果.

基于小样本训练集的肤色模型建立方法

肤色模型的应用大多是建立在小样本训练集上,而目前常用的肤色模型理论却主要是利用大训练集来提高模型的精确性。那么,如何实现一种基于小样本训练集的且具有较高精确度的肤色模型呢?文章在对该领域常用的三种典型肤色模型分析比较的基础上提出了一种多模型相互结合的肤色模型———统计色度空间模型。实验表明在四种模型中效果是最好的。

Characterization of a New Methodology Based on the Intensity of Skin Staining of Ornamental Fish with Applications in Nutrition

The colors presented in ornamental fish determine their acceptability in the market, which is one of the main factors which determine their price. Selecting the fish by only looking at them makes the result analysis based in statistics difficult. Thus, it has become necessary to develop a method of quantification for practical use, which will apply minimal stress to the fish while being economically feasible. The proposed method uses a technique which converts the colors of interest to numerical indexes usable in statistical methods of higher precision. The results increase the abilities of fish nutrition researchers aiming to intensify the color characteristics of fish through selective breeding and genetic engineering with the intent of reliably achieving the desired visual characteristics for commercialization.