对主动轮廓模型而言，初始轮廓尤为重要，本文就初始轮廓提出了一种新的基于“形状共享”思想的初始轮廓学习方法。利用不同种类的物体全局形状或局部形状可能相似的现象，首先提取测试图像的局部形状；再找出样本库中与其局部形状相匹配的局部形状集；根据测试图片与样本图片中局部形状的相对位置及大小，进行全局形状映射；最后依照全局形状的覆盖率分组，融合成一系列初始形状。将这一系列的初始轮廓作为主动轮廓模型的初始迭代函数。另外，本文的主动轮廓模型结合了测试图像的边缘信息与区域信息，利用彩色梯度表示边缘的变化。从实验结果看，将本文学习到的初始轮廓加入混杂主动轮廓中能包含更丰富的形状信息，获得更准确的分割结果，收敛速度更快。

针对经典弹性配准方法在医学图像应用上计算复杂、方向信息不足问题进行了研究。提出一种基于非下采样剪切波变换（NSSTR）的医学图像配准方法。利用剪切波能够对各个尺度、方向和位置实现较好定位的优势。先用该算法对医学图像做非下采样剪切波变换，得到各尺度、方向子带的剪切系数，然后对高频、低频的变换图像分别使用不同的配准方法，最后统一到变化网格上得到最终配准图像。实验结果表明，本方法与传统方法相比，不仅具有配准精度高、鲁棒性好的特点，而且计算效率更高。

在人脸识别中，传统有效的鉴别分析方法需要更多样本评估类内散度信息，由于人脸的单样本问题，导致某些经典的方法，如Fisherface和Eigenface等失效。解决的方法通常是生成各种虚拟样本来扩充训练集来实施这些算法。针对这个问题，根据人脸的对称相似理论，人脸样本的相关变化信息可以从它的对称脸上提取，提出组合原始训练样本及它的虚拟平均脸、对称脸作为训练样本集，应用稀疏理论进行加权积分融合，分两步进行识别的方法。并在ORL和FERET人脸数据库做了对比实验，根据实验结果表明本方法比现有一些突出效果人脸识别方法更好，并具有一定的鲁棒性。

针对完备鉴别局部保持投影算法所求得的最优判别矢量间存在信息冗余问题，提出了核的正交完备鉴别局部保持投影算法。通过将核函数技术与正交性原理融合，采用高斯核函数将原始样本映射到高维特征空间，在高维特征空间的局部总体散度矩阵中计算最优判别矢量，只需在整个范围内对值域空间进行特征值分解。去除局部零空间达到样本降维目的。该算法分别在UMIST人脸库和JAFFE人脸表情库上进行实验，实验结果表明算法的识别率高达95.59%。

In this paper, we apply a Hidden Markov Model (HMM) to analyze time-series personal health checkup data. HMM is widely used for data having continuation and extensibility such as time-series health checkup data. Therefore, using HMM as probabilistic model to model the health checkup data is considered to be suitable, and HMM can express the process of health condition changes of a person. In this paper, a HMM with six states placed in a 2×3 matrix was prepared. We collected training features including the time-series health checkup data. Each feature consists of eight inspection parameters such as BMI, SBP, and TG. The HMM was then built using the training features. In the experiments, we built five HMMs for different gender and age conditions (e.g. male 50's) using thousands of training feature vectors, respectively. Investigating the HMMs we found that the HMMs can model three health risk levels. The models can also represent health transitions or changes, indicating the possibility of estimating the risk of lifestyle-related diseases.

This pilot study tests the effects on individual performance of dyadic versus individual practice in a 3D virtual world (VW) home safety assessment. Sixty medical students in three conditions (dyadic spatially separated with paired avatars DPA; individual with avatar IND; and dyadic spatially together with single avatar DSA) participated in a geriatric home safety simulation. The participants, via avatars, conducted an assessment of physical hazards. Participants then worked individually in a separate 3D VW home assessment. Dyadic practice, spatially together with a single avatar (DSA), improved individual performance in the subsequent 3D VW home assessment.

Our objective was to enable an end-user to create complex queries to drug information sources through functional composition, by creating sequences of functions from application program interfaces (API) to drug terminologies. The development of a functional composition model seeks to link functions from two distinct APIs. An ontology was developed using Protege to model the functions of the RxNorm and NDF-RT APIs by describing the semantics of their input and output. A set of rules were developed to define the interoperable conditions for functional composition. The operational definition of interoperability between function pairs is established by executing the rules on the ontology. We illustrate that the functional composition model supports common use cases, including checking interactions for RxNorm drugs and deploying allergy lists defined in reference to drug properties in NDF-RT. This model supports the RxMix application , an application we developed for enabling complex queries to the RxNorm and NDF-RT APIs.