A Multi-Scale Gradient Algorithm Based on Morphological Operators

Watershed transformation is a powerful morphological tool for image segmentation. However, the performance of the image segmentation methods based on watershed transformation depends largely on the algorithm for computing the gradient of the image to be segmented. In this paper, we present a multi-scale gradient algorithm based on morphological operators for watershed-based image segmentation, with effective handling of both step and blurred edges. We also present an algorithm to eliminate the local minima produced by noise and quantization errors. Experimental results indicate that watershed transformation with the algorithms proposed in this paper produces meaningful segmentations, even without a region-merging step.     

基于DF-OFDM多中继系统功率分配的梯度算法

DF-OFDM技术必将成为现代移动通信的核心技术,但是合理地进行功率分配是必须要解决的问题之一。传统的二分法,大多在精确度和计算量之间很难取舍。在移动通信系统抗信道衰落和提高信道容量方面,协作通信是一种很好的解决方案。针对改进型的多中继DF-OFDM系统,对一种降低计算量的梯度算法进行了简单的介绍与分析,并与二分法进行了比较。仿真结果表明,在功率约束的条件下,梯度法要优于二分法。     

改进的基于梯度幅值的自适应边缘检测算法

分析文献[1]给出的算法.针对其中存在的问题和不足,给出了一个改进算法.实验结果表明,改进算法弥补了原文算法中的缺陷,在获取连续边缘的同时也较好地抑制了噪声,在运算效率方面有了明显的提升,并节省了大量的内存空间.     

基于HOG及在线多实例学习的目标跟踪算法

传统的基于色彩直方图或空间色彩直方图的跟踪算法在跟踪目标出现尺度变化的复杂条件下,因无法显著区分颜色相近的目标和背景,不能得到准确跟踪结果.提出基于HOG及在线多实例学习的目标跟踪算法.此算法采用HOG特征值提取方式,结合在线多实例学习技术,对目标远离场景、平移、旋转、遮挡等情况进行跟踪.实验结果表明,该算法能够对各种复杂情况下的动态目标进行有效跟踪,具有良好的鲁棒性和准确性.     

PROPERTIES OF W/Cu FGMS CONTAINING 1%TiC OR 1%La2O3 PREPARED USING GSUHP

W / Cu functionally gradient materials (FGMs) containing 1%La_2O_3 and 1%TiC were prepared using graded sintering under ultra-high pressure (GSUHP). The specimens have been found to exhibit low porosity (11.57% and 11.35%, respectively). Shearing strength of the specimens between layers is good. Moreover, the specimens have still demonstrated good performance in testing thermal-shock resistance. When power density of laser is 200MWm~(-2), the specimens have been tested for thermal-shock resistance (1000 times); the specimens that contained 1%La_2O_3 were not subjected to damage, whereas those that contained 1%TiC began to crack. Finally, effect of additives on thermal-shock resistance was also preliminarily discussed.     

Albumin Binds Doxorubicin via Self–Assembling Dyes as Specific Polymolecular Ligands

Congo red (CR) type self–assembled ribbon–like structures (SRLS) were previously shown to interact with some proteins, including albumin. SRLS also complex with some drugs with a flat, ring–shaped structure with aromatic characteristics, intercalating them into their ribbon structure. The combination of interaction with proteins and drug binding by SRLS enables the use of such systems for immunotargeting. It is especially interesting in the case of chemotherapeutic agents. The present experiments aimed to show that the model carrier system composed of supramolecular albumin and Congo red efficiently binds doxorubicin (Dox) and that the drug can be released at reduced pH. The presented results come from the studies on such complexes differing in the molar ratio of CR to Dox. The following methods were used for the analysis: electrophoresis, dialysis, gel filtration, spectral analysis, and analysis of the size of the hydrodynamic radius using the dynamic light scattering method (DLS). The applied methods confirmed the formation of the CR–Dox complex, with large dimensions and changed properties compared with free CR. The presented results show that albumin binds both CR and its complex with Dox. Various CR–Dox molar ratios, 5:1, 2:1, and 1:1, were analyzed. The confirmation of the possibility of releasing the drug from the carriers thus formed was also obtained. The presented research is important due to the search for optimal solutions for the use of SRLS in drug immunotargeting, with particular emphasis on chemotherapeutic agents.     

混合梯度稀疏先验约束下的图像盲复原

图像盲复原旨在无参考的情况下准确估计模糊核并恢复潜在的清晰图像。现有研究成果表明,利用全变分模型对高阶图像梯度先验约束进行描述可以有效抑制复原图像中产生的阶梯效应。本文在实验观察和研究的基础上,提出了采用稀疏先验约束模型对盲复原过程进行正则化的方法,以获得更佳的图像复原效果。该方法利用图像高阶梯度的稀疏性,通过与低阶梯度相结合来构造混合梯度正则项。同时,在正则项中引入基于图像熵的自适应因子,来调节迭代优化过程中两类梯度先验的比例,以此获得更好的收敛性。仿真与实验证明,与现有图像盲复原先进方法相比,本文方法具有更优越的图像复原性能。     

Böhler Edelstahl

Böhler Edelstahl produces high performance tool and high speed steels via an optimised powder metallurgy/hot isostatic pressing route, as part of a high value special alloys portfolio.     

负压差对固体进料系统压力分布和气固两相流动规律的影响

A series of experiments on a solid feed system was performed to investigate the effects of negative pressure gradient on the gas-solid flow pattern and hydrodynamic characteristics. Based on the non-fluidized gas-solid two phase flow and particulate mechanics in the standpipe, a method for predicting the pressure of the air passing through the recycle chamber and the pressure drop through the loop seal slit in these systems is also presented. The predicted pressure profile along the negative pressure gradient from the proposed model exhibits a good agreement with the experimental data. The experimental data show that the gas flow in the standpipe is always upward in the negative pressure gradients, while the direction of gas cannot be changed by the drag force. The solid flow rate, the materiel bed height in the recycle chamber (Hf), the pressure drop across the slit (△Pg） decrease with an increasing negative pressure gradient. At a given negative pressure gradient, the superficial gas velocity through the recycle chamber of the loop seal does not affect the pressure drop in standpipe. It increases with an increasing negative pressure gradient.     

基于改进Canny算子的垃圾图像边缘检测

为提高垃圾识别分类的准确率,文中在垃圾图像预处理过程中提出了一种基于改进Canny算子的垃圾图像边缘检测方法。该方法从传统Canny算子滤波方式、梯度方向及阈值自适应3个角度实现了垃圾图像边缘检测的优化。针对Canny算子高斯滤波仅适用于高斯噪声和边缘细节易丢失的问题,采用改进的梯度倒数加权法进行滤波。针对Canny算子易检测出伪边缘的问题,通过在计算图像梯度方向的过程中增加方向梯度模板实现了边缘的精确化。同时采用最小误差法解决人工设定阈值的局限性,实现阈值自适应。实验结果表明,该方法在去噪性能和边缘细节两方面得到了改进,获得了更好的边缘检测效果,为后续垃圾图像的识别分类提供了技术保障。