图像

Brillview Technology, a leader in Chinese image processing sector, has created pure hardware image processing techniques, which are not only innovative and originative but also help solve various challenges in image processing applications such as multiple RGB signals display, free overlapping of different types of signals, acquisition and output of ultra-high resolution signals, computer collapse and virus corruption, etc. This enterprise advances Chinese image processing technology up to an international advanced level and opens a new era of display control.

作为中国市场图像处理领域的领导者，博睿科技独创的纯硬件的图像处理技术，极具创新和想象力，解决了图像处理领域RGB多信号显示，各种信号类型之间的自由叠加，超高分辨率信号的采集和输出，死机和病毒困扰等诸多瓶颈，带领国内的图像处理技术进入国际先进水平，开创了显示控制的新纪元。

Looking upon the patches of face images as some specific class of signals, their eigen-subspace can be found by Principal Components Analysis, and the super-resolved facial patches are regularized in the orthogonal complement of the eigen-subspace. This eigen-subspace based regularization is combined with the classic reconstruction constraint under a Bayesian framework to produce the high-resolution outcome.

将人脸图像块看作一些特定信号类，利用主成分分析计算训练人脸图像块的特征子空间；并将传统的"重建约束"与人脸图像块的"正交补特征子空间约束"统一在贝叶斯框架下，提出一种人脸图像SR的规整化方法。

Compare the complexity and precision of the polynomial method and orthogonal polynomials method used in geometrical adjustment, an experiment is made; some surface fitting method used in spatial co-registration of multiresolution image are studied and experiments are made to show their different in effects; in order to increase the spectral inherit, we make a histogram match in images to be fused.

比较了图像几何校正的多项式方法和正交多项式方法在复杂度与精度上的差异，进行了仿真实验；研究了不同插值技术在图像配准中的应用，对不同插值方法的效果进行了比较，给出了示意图；为消除图像灰度级不匹配对融合结果的影响，对待融合图像的灰度直方图进行了匹配处理。

Accordingly pels tracing and function approximation are avoided ,and vectorization process is simplified effectively.At first ,every arithmetic of image transformation and image filter is researched and its disposal image is analyzed .

本文首先对图像变换和图像滤波的各种算法进行了研究，分析了各个算法的处理效果，根据实验试验结果提出应结合图像本身的特点选择合适的预处理方法，并运用小波变换对图像进行压缩变换和滤波处理。

Preoperatively and postoperatively computed tomographic scanning was performed in every patient. For the CAD treatment group, firstly, the 3D graph of the patient's skull was built based on the CT data using the reconstructing function of Mimics 8.11. Then the 3D file was imported to Geomagic 4.0. The injury lateral was resected and replaced by the mirror of the intact side, so an intact skull was generated at last. Consequently, a model was manufactured in a technical machine. The last step was to prebend the titanium mesh on the model.

CAD/CAM治疗组是利用术前诊断CT，将DICOM格式CT数据输入图像处理软件Mimics8.11，由软件生成颅颌面骨STL格式的三维图像文件，应用另一个图像处理软件Geomagic4.0读取该文件，并应用镜像工具以中线为轴线生成健侧半的镜像，最终得到双侧对称的颅颌面模型，将该图像信息输入快速成型系统，通过逐层树脂堆积，生成树脂头模。

In this paper, we research the feasibility of using the Walsh and Ranktransformation characters as match primitivity in replace of gray character of image. Theexperiment results indicate that the approach is feasible and got a better result comparedwith using grey character as primitivity.

矩理论在图像分析和处理方面有着广泛的应用，从数字图像计算出来的矩集，通常描述了该图像的全局特征，并提供了大量的关于该图像不同类型的几何特性信息。

We make use of the ideas of simulated annealing and Powell optimization to handle the relationship of 3D structure and 2D image and to reconstruct the projection matrix.

我们采用了基于模拟退火和Powell优化思想的方法处理二维图像和CT图像的结构信息，重建出在二维图像和CT图像数据之间的变换矩阵。

A novel multisensor images fusion algorithm is presented based on lifting quincunx wavelt. First, the lifting quincunx wavelet transform is used to perform a multiscale decomposition of each image. Then, the wavelet coefficients of fused image are constructed using multiple operators according to different fusion rules. Finally, the reconstructed image is obtained through the inverse transform.

提出了一种基於提升方案的五株形小波的多传感器图像融合方法，即在提升五株形小波多尺度分解的基础上，对每一幅图像进行多尺度分解，按照不同的融合规则和融合运算子去构造融合图像对应的各小波系数，再通过逆变换重构融合后的图像。

This paper discussed its theory mechanism at terms of radiancy. Also, the results of visual interpretation and quantitative analysis to gray variance index show that the Low-pass RF transform have a capability of retaining color from multi-spectral image.

该融合法首先对高几何分辨率的全色波段进行低通滤波，然后将低分辨率多光谱图像与全色波段图像相乘，再除以滤波后的全色波段图像，便得到融合图像。

The images of ranbutan were obtained by CCD camera. Image processing algorithm of OSTU and area labeling were applied to 48 rambutan images in order to remove the images of background and stem, and the texture features of color and lustre were extracted. A multi-grade support vector machine was set up, and color and lustre features were the inputs of DAGSVM.

通过CCD采集红毛丹可见光图像，经OSTU分割算法来分割图像背景后，采用面积标记算法得到去除长穗梗区域的红毛丹图像；然后提取基于色度的红毛丹图像的彩色纹理特征，并用多分类支持向量机的模式识别方法来识别红毛丹色泽等级。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。