glatiagf的博客

跨座式轨道梁表面出现的裂纹和缺陷，严重影响了列车安全及平稳运行。由于采集的图像存在边缘模糊、局部或整体 对比度差等不足，提出一种基于曲波变换的梁面裂纹增强新算法。根据裂纹图像曲波变换后的不同尺度系数，由于分数阶微分能增强信号的中频成分、非线性保留信号低频的特点，在曲波域的低频采用分数阶微分增强纹理信息; 同时在曲波域的方向子带根据方向的敏感性，采用对弱边缘增强、强边缘保留、噪声去除的处理。并对低频和方向子带增强时的参数分别进行了讨论。实验结果表明，所提出的方法与其他算法相比，主观上细节增强和对比度提高均优于其他算法，客观上对比度改善指数( contrastimprovement index，CII) 也大大提高，表明提出的算法增强效果良好。

提出了一种融合分水岭变换和图论的图像分割方法。利用图像的局部灰度信息进行分水岭变换后, 将图像分割成多个小区域, 再结合各小区域的灰度和空间信息从全局角度用Norm a lized Cut方法在区域之间进行分割, 产生最终的分割结。实验结果表明, 该方法可以消除分水岭变换后所产生的过分割现象, 是一种有效的图像自动分割方法。

为了满足图像识别较高的识别率和较低的时间消耗两方面的要求, 提出了一种不变矩和禁忌搜索结合算法, 即以不变矩作为特征参数, 以禁忌搜索算法加快图像数据检测。在最佳的迭代步长下, 不变矩和禁忌搜索结合算法在平均识别率、消耗时间、平均重复个数等参数方面, 要明显优于一般不变矩算法。不变矩和禁忌搜索结合算法兼顾了识别率和时间消耗两个方面的因素, 可以应用于一些高速的图像识别场合。

本文针对玉米种子图像识别的特点, 提出了基于模糊理论的玉米种子图像识别方法。文中将禁忌搜索算法(Tabu search)引入到C- 均值聚类中, 以克服模糊C- 均值聚类的局部性和对初始聚类中心的敏感性; 最后, 详细设计了上述各算法。该方法为种子的在线检测提供了一种新思路, 也拓展了模糊理论的应用范围。

详细介绍了制造爬行机器人的结构和控制系统

Developments in measuring surface temperatures during grinding at high specific material removal rates using PVD low melting point coatings are reported. This paper considers the use of this type of coating to determine the finished surface temperature and the thermal gradient below the workpiece surface under aggressive grinding conditions. Theoretical surface temperatures derived from mathematical modelling are compared to those measured by the PVD coating technique. Theoretical temperatures showed good correlation to the measured temperatures. Metallographic examination including Vickers microhardness testing was used to validate the measured thermal gradients and depth of phase transformations. The effect of the heating rate due to the grinding conditions was shown to have a significant influence on the A3 austenitic transition temperature. The technique of using low melting point PVD coatings is shown to offer a robust measurement technique for the determination of the finished surface temperature of a ground component, within aggressive environments and at high specific material removal rates.

An automatic grinding system with grinding force control on a CNC machining center is developed to reduce surface roughness. This system includes an electric hand grinder mounted on a CNC machining center, a force sensor to measure the normal grinding force, and a force control sub-system to adjust the grinding depth. The CNC spindle vertical movement is produced by “external machine zero point shift”. The experiments performed in this study include: (1) Taguchi’s parametric design to determine an ideal combination of the horizontal feed rate and the desired force value during force control and (2) grinding along an inclined surface. The results show that the force control technique may indeed reduce the average grinding force and grinding force variation.