动态分数选择的多视角非接触指纹识别融合

与传统的二维指纹采集方式相比,非接触的指纹采集方式可以提供更丰富的特征,并能克服二维采集方式某些不足．基于多视角非接触指纹识别融合问题,提出了基于聚类的动态分数选择(CDSS)算法．首先,通过聚类将匹配分数分为两类,分别计算两类中匹配分数的个数以及其他统计量,然后通过参数的判断选择相应的统计量作为整个系统最终用于决策的匹配分数．实验表明:相比单视角非接触二维指纹识别,基于CDSS的多视角非接触指纹识别的识别性能有显著提高．与sum、max、SVM和Fisher线性判别算法等的实验比较也验证了CDSS在多视角非接触指纹识别融合方面的优越性．     

高温转速传感器的研制

介绍了非接触磁电式转速传感器的基本原理、结构设计及磁路计算,给出了试验数据,并对测试结果进行了分析。讨论了高磁性材料在高温环境下的应用。     

非接触指纹图像识别算法研究

非接触指纹识别具有接受程度高、防伪性高、卫生性高等优点,是目前生物特征识别领域的研究热点,但是非接触指纹图像的背景区域比接触式的相对复杂,指纹图像会出现旋转和平移现象,且脊、谷线的对比度较低,这些因素严重影响了非接触指纹的识别性能.采用接触式指纹图像处理方法很难达到良好的处理效果.本文根据非接触指纹图像的特点提出了相应的非接触指纹图像的预处理方法.首先采用图像YCbCr模型中的Cb分量和Otsu法相结合的方法进行手指区域的提取.其次采用高频强调滤波和迭代自适应直方图均衡化相结合的图像增强算法进行图像增强处理,再用简化的Gabor函数模板进行二次增强,然后提出了一种手指指纹ROI区域提取的方法.最后本文采用基于AR-LBP算法进行特征提取,利用最近邻分类器进行特征匹配.实验结果表明,本文提出的非接触指纹算法能够达到很好的图像识别效果.     

Head Movement Based Interaction in Mobility

The touchless techniques in human computer interaction (HCI) can effectively expand communication capabilities. In the paper we present the innovative touchless computer control method based on head movement analysis. The aim of our work was to replace the standard mouse with the movements of the user’s head. In contrast to the known solutions, our proposition does not require image recording of the user’s head and complex image analysis. The analysis of position in our solution is made using the camera worn by the user on the head. A project of such a solution has been developed and the research of it has been carried out. It has been shown that in this way it is possible to effectively move the screen cursor to the position which is identified by the user’s face orientation. Additionally, in this solution, the eye image analysis has been performed. Interpretation of blinking allowed executing system commands. Using the built prototype the experiments have been carried out in a group of 30 people. Studies have shown high efficiency and ergonomics of the proposed solution.     

A Dual-Responsive Artificial Skin for Tactile and Touchless Interfaces

The progress from intelligent interactions and supplemented/augmented reality requires artificial skins to shift from the single-functional tactile paradigm. Dual-responsive sensors that can both detect pre-contact proximal events and tactile pressure levels enrich the perception dimensions and deliver additional cognitive information. Previous dual-responsive sensors show very limited utilizations only in proximity perception or approaching switches. Whereas, the approaching inputs from the environment should be able to convey more valuable messages. Herein, a flexible iontronic dual-responsive artificial skin is present. The artificial skin is sensitive to external object's applied pressure as well as its approaching, and can elicit information of target material categories encoded in the proximal inputs. Versatile applications are then demonstrated. Dual-mode human–machine interfaces are developed based on the devices, including a manipulation of virtual game characters, navigation and zooming in of electronic maps, and scrolling through electronic documents. More importantly, the proof-of-concept application of an entirely touchless material classification system is demonstrated. Three types of materials (metals, polymers, and human skins) are classified and predicted accurately. These features of the artificial skin make it highly promising for next-generation smart engineered electronics.