Design,simulation, and implementation of an integrated,hybrid photocurrent-to-digital converter in CMOS technology

This article presents a design, modeling, simulation, and measurements of a hybrid photocurrent-to-digital converter integrated together with photo-diodes in a 130 nm CMOS process, without any additional process steps. Photo-currents of integrated photo-diodes with different responsivities to different wavelength of the light and light intensity are converted into 22-bits digital results in 2 ms. The results can then be converted into CIE XYZ or RGB color luminosity space using dedicated DSP algorithm. A high resolution, hybrid, ADC converts light induced photo-currents into 22-bit digital results, canceling the dark current of the photodiodes and 1/f noise and offset voltage of the input stage of the modulator. The whole converter consumes on average less than current at supply voltage at 10 conversions per second. It occupies approx. 0.8 mm² of silicon area, including the three photodiodes, a multiplexer, and the ΣΔ modulator.     

不同调光方式LED照射下的人体非视觉效应

分别采用模拟调光和PWM调光方式实现LED光调节。采用生理参数法,以在校大学生为研究对象,选择3种色温(3 000,5 600,6 500 K)LED光源,在3种照度(300,500,800 lx)水平下,对8名(5男,3女)受测者在实验前后的脉搏和瞳孔进行测量。选择同样3种色温LED光源,分别采用两种调光方式添加蓝光(464nm,20 lx)实现300 lx的照明环境,对12名(8男,4女)受测者在实验前后的脉搏和瞳孔进行测量。结果表明:两种调光方式实现的LED光环境对人体脉搏变化率和瞳孔收缩率的影响均没有显著性差异;在测试照明环境中添加蓝光,可引起人体脉搏变化率和瞳孔收缩率增大,峰值处分别增加约6%和9%。     

Flux and fluence dependence of disorder produced during implantation of 11B in silicon

The flux and fluence dependence of disorder produced in silicon during the implantation of ¹¹B has been investigated at room temperature, -50°C. and -120°C. Implantations were carried out with 200 keV ¹¹B ions using current densities in the range from 0.06μA/cm² to 15μA/cm², and the disorder monitored by measuring the energy spectra of backscattered protons which were incident on the sample at 450 keV parallel to a (110) axis. Significant differences in the dependence of the disorder on ¹¹B flux and fluence were observed between the implantations performed at room temperature and those carried out at the two lower temperatures.     

Color centers in irradiated MgF2

Abstract

Color centers in rutile-structured MgF₂ single crystals irradiated at 20K/360K by reactor neutrons are investigated spectroscopically at LNT. Four different types of the F-F vacancy bond in MgF₂ are possibly identified to the observed absorption bands as due to the F₂ centers; instead of previous assignments, the 300nm band to the F₂(D_2h), the 325nm band to the F₂(C₁), the 355nm band to the F₂(C_2v), and the 400nm band to the F₂(C_2h) centers.     

机动车驾驶人场地驾驶技能考试系统算法设计

目前机动车驾驶人场地驾驶技能考试都是采用车载工控机进行软件算法的实现,其成本高,要求量取较多的车辆尺寸,且需要对场地项目进行区域划分,算法效率低。针对以上问题,设计了基于RTK技术的差分双GPS定位方案的车身出线及车辆压线的高效评判算法,该算法主要采用线段相交式的计算方法,通过实际场地测试,得出该算法对车辆尺寸度量点数少、运算效率高,易于在微控制器中实现,成本低,且无需对场地项目进行GPS区域划分,可以缩短施工周期、提高施工效率。     

高效的基于段模式的恶意URL检测方法

提出一种高效的基于段模式的检测恶意URL的方法,该方法首先解析已标注的恶意URL中的域名、路径名和文件名3个语义段,然后通过建立以三元组为词项的倒排索引快速计算恶意URL每个语义段的模式,最后基于倒排索引查找到的段模式来判定给定的URL是否是恶意URL。不仅如此,该方法还支持基于Jaccard的随机域名识别技术来判定包含随机域名的恶意URL。实验结果表明,与当前先进的基准方法相比,该方法具有较好的性能和可扩展性。     

多摄像机在无重叠区域的目标匹配算法

基于颜色特征的目标匹配方法具有很好的区分性和直观性,但是特征本身受光线变化、背景复杂度等影响而对于目标识别准确率存在差异.针对这一现状,提出了一种新的基于时空局部特征和颜色特征的目标匹配方法.利用时空局部特征3-D LSK对待测视频中的运动目标进行行为识别,识别出具有相同行为目标.再结合目标的颜色特征,通过基于区域权重(Weighted Region Matching)的颜色匹配方法来识别出具有相同颜色的目标.对无重叠区域的多个测试视频进行的相关实验表明,该算法能够有效识别出普通场景下的不同目标,对于衣着颜色相近的不同目标也有很高的识别率.     

应用于CMOS图像传感器的联合降噪插值算法研究

提出了一种可应用于CMOS图像传感器的联合并行图像预处理算法,其中,色彩插值算法采用双线性插值法和边缘导向插值法相结合的方式来完成色彩还原,而降噪算法采用带阈值的空间自适应滤波降噪算法,在硬件语言描述算法时,将两种算法的处理过程进行了联合处理.实验仿真结果表明,通过对算法进行ISE仿真和Matlab显示处理,得到色彩饱和度较高的彩色图像,峰值信噪比(PSNR)为30～36 dB.整个处理过程有效地简化了片上系统(SOC)的图像处理过程,提高了算法的运算速度,节省了芯片面积且降低了系统功耗,为CMOS图像传感器的片上系统集成化设计提供了技术支持.     

基于肤色模型和改进Adaboost算法的人脸检测

为提高人脸检测的精度,提出一种基于YCbCr肤色模型和Adaboost算法的人脸检测方法。利用YCbCr颜色空间的肤色阈值模型确定待检测区域。再利用Adaboost算法准确定位出人脸位置,并对Adaboost算法的训练过程提出一种新的权值更新方法,防止权值过分增大,避免了Adaboost算法训练过程中出现的“退化”现象。