改进LDP结合几何特征融合的人脸表情识别

通过局部方向模式(LDP)提取的局部外观特征常用于人脸表情识别中,但其存在容易受噪声像素影响和伪编码的问题。为此,提出一种改进型的LDP编码方案(ILDP)。利用Sobel算子代替Kirsch掩模来提取图像中的梯度信息,将梯度信息进行对数变换后进行累加,避免噪声像素点的影响;通过采用一个梯度幅度阈值判断是否为平坦区域,避免进行错误的LDP编码。为进一步提高识别率,将ILDP提取的局部外观特征与主动表观模型(AAM)提取的全局几何特征相结合,通过PCA进行降维。在JAFFE和BU-3DFE两个公共数据集上的实验结果表明,该方法能够有效提高表情的识别率,降低对噪声的敏感性。

改进的Mallat算法在金属地震数据降噪处理中的应用

野外地震数据所处地质环境复杂,受干扰信息影响严重,信噪比低、分辨率低等,不利于地质数据解释及后期地质勘探,单一的降噪算法很难有效去除各种干扰信号。本文提出一种改进的Mallat算法对地震数据进行降噪处理,利用Mallat算法将地震信号分解到不同尺度频带空间上,再选取合理的阈值对不同频带空间上的地震数据进行降噪处理,然后利用ICA分离方法在不同尺度频带空间上提取有效信息,通过Mallat算法将有效信号重构回原始地震信息。用构建的Mallat算法对实际金属地震数据进行降噪处理,结果表明,改进的Mallat算法降噪效果明显,能较好地去除地震数据的各类干扰,降噪后的地震剖面纹理清晰、分辨率高。

The improved ICA algorithm and its application in the seismic data denoising

The field seismic data is disturbed by the interferential information, which has low signal to noise ratio (SNR). That is disadvantage for seismic data interpretation. So it is important to remove the noise of seismic data. Independent component analysis (ICA) can remove most of the noise interference. However, ICA has some defects in noise reduction, because it needs some conditions that seismic data is independent reciprocally for denoising. To solve these defects, this paper proposes an improved ICA algorithm to noise reduction. Through simulation experiments, it can be obtained that the best decomposition levels of the new algorithm is 3. At last, the proposed improved ICA is applied to deal with the actual seismic data. The results show that it can effectively eliminate most of seismic noise such as random noise, linear interference, surface waves, and so on. The improved ICA is not only easy to denoising, but also has excellent mathematical theoretical properties.

利用开采解放层治理强矿压

针对大同矿区特厚煤层综放工作面在回采过程中出现的强矿压问题,以塔山矿一盘区C3-5号层8108工作面为工程实例,提出开采解放层泄压的方法,通过先行开采8108工作面上覆山4号煤层,使下部综放工作面开采过程中矿压显现强烈的现象得到了有效缓解,保障了特厚煤层综放工作面的安全开采,取得了良好的经济技术效果,为同类型矿井提供了借鉴。