分布式环境照度监测系统的研究与实现

环境光照与人们的生活生产密切相关。为准确测量光照强度,提出了一种分布式环境照度监测系统。通过各采集终端对环境照度进行检测,利用嵌入式TCP/IP技术将数据实时传送给监测中心进行集中处理与分析,经J2EE应用系统进行Web发布,供有权限的用户通过浏览器进行照度查询与管理。该系统实现了照度信息的远程采集、集中处理与网络共享。     

基于DDGVFSnake和Gamma方法处理人脸光照不均

使用一种动态方向梯度矢量流Snake (DDGVF Snake)算法来检测未知图像的阴影和高光区域,根据已知的标准人脸图像库中图像的灰度分布,计算出灰度中值,然后采用加权非线性Gamma灰度矫正算法来归一化待识别图像的光照变化,使之与已知的标准库中的图像灰度一致。在Yale B,CMU PIE和CAS PEAL人脸库上的实验表明,所提出的方法与同类的处理方法相比,具有更高的恢复精度。     

Perceived illumination measured

We consider a method of measurement of apparent illumination using the perceptual phenomenon of fluorescence to which we give a psychophysical explanation based on the hypothesis of direct relation of perception of self-luminosity to the apparent illumination of the scene observed by the viewer. We report results of three experimental studies: (1) measurements of apparent illumination when a flat mondrian in the darkness is observed; (2) measurements of apparent illumination in a “rich” 3D scene; (3) testing linearity of the subjective illumination space. Based on the obtained data, we confirm the basic hypothesis of a relation between the self-luminosity phenomenon and the perceived illumination, and show that the suggested method of measurement of apparent illumination gives reliable and accurate (errors were in the range 4–10%) values of its chromatic components. Apparent illumination in the case of a mondrian observed in the darkness is found unstable, which can be expected given that the context of such a scene is insufficient for obtaining a unique solution for all the components of the photometric problem. In the second experiment, we show that the measurement results remain constant when the context of a “rich” 3D scene is changed and that they follow the changes in illumination conditions. In the third experiment, we study the composition of different illuminations and show that the simple relation f = f₁ + f₂ holds for the measurement of the illumination (f) when two other illuminations (f₁, f₂) are applied simultaneously. © 1998 John Wiley & Sons, Inc. Col Res Appl, 23, 159–168, 1998     

Priority-driven Ray Tracing

A typical ray tracing algorithm traces a ray through each screen-pixel and spawns secondary rays at ray–object intersection points. Unlike traditional ray tracers which follow these rays recursively, we assign a priority value to each newly spawned ray and insert it into a priority queue. The priority assigned to each ray can be based on a variety of criteria, some of which we explore here. The next ray we trace is always the one with the highest priority in the queue. Occasionally, we trigger display updates when a checkpoint or predefined threshold is reached, providing intermediate images for review and evaluation. Classical ray tracers, once given the rendering specifications, are not controllable by the user. The priority-driven ray tracing, on the other hand, provides the user with a mechanism to steer the rendering and deliver intermediate images amid processing. This paper describes the illumination model of the non-recursive priority-driven ray tracer and evaluates its memory and time requirements. We show that although worst-case memory requirements can be overwhelming, in practice, our method is both useful and feasible. © 1997 by John Wiley & Sons, Ltd.     

融合多分支残差与仿射变换的低照度图像增强

针对现有低照度图像增强算法难以同时处理亮度、对比度、伪影和噪声等因素,提出了多分支残差与仿射变换低光增强网络,其核心思想是运用不同模块处理不同的任务.首先通过光照估计模块获得低光图像的光照变量,然后使光照仿射变换模块与光照编码参数融合恢复图像的光照,最后通过细节重建模块融合更多的图像细节获得最终输出.实验结果表明,该方法有效地丰富了图像的纹理细节,同时增强了亮度和对比度,并具有更少的伪影和噪声.通过与其他主流方法进行比较,定量和定性地证明了提出方法对低光图像增强的效果更好.     

Book of the Quarter: Managing Global Innovention

Book reviewed: R. Boutellier, O. Gassman and M. von Zedtwitz, Managing global innovation: Uncovering the secrets of future competitiveness     

基于平滑度测量与互相关制约的图像修复算法

目的为了解决当前较多图像修复算法在对损坏区域实施填充修复时没有考虑图像块之间的相关性,导致修复图像中存在振铃以及连接间断等不足,拟设计平滑度测量因子耦合互相关制约的图像修复算法。方法将图像的Laplace算子引入到优先权的计算过程中,以增加图像的边缘信息,优化数据项的成分,并利用置信度、图像的Laplace算子和数据项构造优先权计算模型,以获取优先修复块。利用等照度线的二阶导数,构造平滑度测量因子获取图像的平滑度,并以图像的平滑度为导向,对最优匹配块的搜索区域进行定位。通过误差平方和函数(SumofSquaredDifferences,SSD)在定位的搜索区域中搜寻最优匹配块,并采用互相关系数函数对最优匹配块的唯一性进行制约,以提高所获最优匹配块的准确性。结果实验结果显示,与当前修复技术相比,所提方法具有更高的连接完整性和清晰度。结论所提方案可对损坏图像完成较好的视觉复原,在图像信息处理领域具有一定的参考价值。     

一种基于平面拟合的图像恢复方法

论述了在进行实际图像处理时所使用的一种对于不均匀光照图像进行恢复的方法。该方法基于平面拟合来估算出实际的光照强度分布,并根据图像各点的反射系数不变这一特性来对图像进行修正。实验结果表明,该方法有效地克服了不均匀光照所造成的对于图像处理的不利影响。     

大功率白光发光二极管光源的制作

采用GaN基蓝色发光芯片为激发源,结合黄色硅酸盐系列荧光粉封装成大功率白光发光二极管(W-LEDs).利用24颗大功率5W白光发光二极管制作了两种不同连接方式的W-LEDs路灯:2并12串,和4并6串.设计了相应的驱动电路,对这两种不同连接方式的大功率W-LEDs路灯的光电特性及其在照明光源中的应用条件作了深入地研究和对比,测试了它们的伏安特性,发光效率以及功效,结果表明2并12串连接方式的W-LEDs路灯具有更加稳定的伏安特性,更高的照度以及更高的功效.与高压钠灯和荧光灯的特性相比较,W-LEDs路灯作为绿色环保光源灯,具有更高的显色指数,更加环保,节能.     

LTP与光照突变补偿结合的运动目标检测算法

针对现有运动目标检测算法在光照突变条件下鲁棒性不强、易发生误检等问题,提出了将LTP算子与光照突变补偿模型结合的运动目标检测算法。首先利用LTP算子获取当前帧与其背景图像的纹理特征图,然后计算当前帧图像像素点作为前景点的概率并依据概率自适应地更新背景,再判断图像是否发生光照突变补偿并采用线性模型补偿突变图像序列,最后利用背景减除法获取运动目标。实验结果表明本文算法在光照突变条件下对运动目标检测效果理想,检测精度指标PR高于同类其他算法,且具有良好的鲁棒性。