TP2

用主成分分析来降维，用Kmeans方法聚类；应用于变化检测中(Principal component analysis is used to reduce dimension with k-means algorithm is used to cluster; change detection)

中国科学院长春光学精密机械与物理研究所 基于外极线约束条件的视频判读方法(Changchun Institute of Optics, Fine Mechanics and Physics epipolar constraint based on the video interpretation method)

图像处理（图像融合，图像加密等）的性能评价指标集包括：平均梯度，边缘强度，信息熵，灰度均值，标准差(均方差MSE)，均方根误差，峰值信噪比（psnr），空间频率(sf)，图像清晰度，互信息(mi)，结构相似性(ssim)，交叉熵(cross entropy)，相对标准差。(Image processing，average gradient, edge strength, information entropy, gray mean, standard deviation (variance (MSE), are root mean square error, peak signal to noise ratio (PSNR), spatial frequency (SF), the sharpness of the image, the mutual information (MI), similar to the structure of (SSIM) and cross entropy (cross entropy). The relative standard deviation)

对图像矩阵进行zigzag扫描,从而达到对图像纹理（灰度值）置乱的效果。( scanning of the image matrix by Zigzag，To achieve the scrambling effect of the image texture (gray value).)

说明：  compression压缩感知算法用于图像处理。信号压缩(compressionCompressed sensing algorithm)

鼠标写毛笔字源码例程，程序结合易语言扩展界面支持库和图像格式转换支持库，实现用鼠标写出毛笔字的效果。可以展现多种毛笔字形态，启动时有使用帮助说明。(Mouse calligraphy source routines, programs combined with easy language support library expansion interface and image format conversion support library , use the mouse to write calligraphy achieve results. )

维纳滤波对图像模糊的处理以及对加入噪声的图像进行清晰处理(Wiener filter for image processing and fuzzy image on a clear noise added deal)

遥感影像变化检测算法(IR-MAD、MAD、CVA、PCA)(change detection method(IR-MAD MAD CVA PCA))

c#编写的图像平滑处理，源码比较全，可以成功编译，希望对大家有帮助。(c# prepared by the image smoothing, source code comparison, can be successfully compiled, we want to help.)

针对多聚焦图像融合，本文首先介绍了小波变换融合算法，虽然小波变换方法无冗余，高频分量损失较小，具有较好的融合效果，但存在两个主要缺点：一是移变性，二是融合图像在重构时会受到一些外来因素的影响，所以仍会丢失一些边缘信息，为了克服小波变换法的以上缺点，可以采用基于小波分解和点扩展函数模型PSF相结合的融合方法。该方法首先对不同的图像进行无降2采样的小波分解，以保持与源图像大小相同，然后分别将其各方向、各尺度的高频分量进行叠加，并依此提取其高频分量的特征信息来判定清晰目标或模糊目标，以此来设计融合算法，采用一定的策略将两副源图像进行融合，从而避开了小波的反变换，因此融合后的图像具有比小波变换法更好的融合效果。(multi-focus image fusion, this paper introduces the wavelet transform algorithm, Although the wavelet transform non-redundant, the loss of high-frequency components are smaller, better integration effect, But there are two major drawbacks : First, the shift variability and the second is the integration of image reconstruction will be subject to a number of external factors. Therefore, there would still be some marginal loss of information, in order to overcome the wavelet transform of the above shortcomings, can be used based on wavelet decomposition and the point spread function PSF model combining the fusion method. The first pair of different images ratchet two sampling wavelet decomposition, and to maintain the same source image size. Then all of its direction, the scale of high-freque)