1329572013522_0

说明：  语义分割，训练样本，模型训练，成果展示，深度卷积神经网络(Semantic segmentation, training samples, model training, achievement display, deep convolution neural network)

说明：  本课题的主要内容是图像预处理,它主要从摄像头中获取人脸图像然后进行处理,以便提高定位和识别的准确率.该模块主要包含光线补偿、图像灰度化、高斯平滑、均衡直方图、图像对比度增强，图像预处理模块在整个系统中起着极其关键的作用，图像处理的好坏直接影响着后面的定位和识别工作,内有源代码和全部论文资料(this issue is the major content of image preprocessing, mainly from the camera to obtain images Face then, in order to improve the recognition and positioning accuracy. The module consists mainly of light compensation, Grayhound, Gaussian smoothing, balanced histogram, image contrast enhancement, image pre- processing module in the system plays a crucial role in image processing will have a direct impact behind the positioning and identification, within Active code and all papers information)

图像拼接，用于合成全景图像，VC++开发环境，C++实现。(Image Stitching)

计算机图形学 c语言实现的直线段裁剪算法(Computer graphics c language to achieve a straight line segment clipping algorithm)

对脑部ct图像的头骨形状进行区域生长算法的运算(Ct images of the brain the skull shape of region growing algorithm for computing)

实现shift算法，进行精确的图像识别，可以抵抗各种旋转、平移、缩放等仿射变换的攻击(Realize shift algorithm, an accurate image recognition, you can resist all kinds of rotation, translation, scaling affine transformation attack)

基于分片重排列技术的维数约简方法，用于特征提取。(Patch Alignment for Dimensionality Reduction，IEEETrans.，2009.)

冲击噪声识别和消除的新算法，可以消除高达95 的冲击噪声，可用于毫米波成像等高脉冲噪声的恶劣环境(A novelt impulse noise removal algorithm which can work under the circumustance that the noise ratio reaches to 95 . )

说明：  CFAR目标跟踪用于雷达目标检测，算法都实测过，可用。(CFAR target tracking)

利用块稀疏的方法求解感知矩阵，在压缩感知、图像恢复方面都可以应用(Method for solving block sparse perception matrix in compressed sensing, image restoration can be applied)