imagematlab

能量检测法是一种比较简单的信号检测方法[1]， 属于信号的非相干的检测， 直接对时域信号采样值求模，然后平方即可得到；或利用FFT转换到频域，然后对频域信号求模平方也可得到。它无需知道检测信号的任何先验知识，对信号类型也不作限制。 实际上，能量检测是在一定频带范围内作能量积累，如果积累的能量高于一 定的门限，则说明信号的存在；如果低于一定的门限，则说明仅有噪声。能量检 测的出发点是信号加噪声的能量大于噪声的能量。能量检测方法对信号没有作任何假设，是一种盲检算法。 此代码“利用FFT转换到频域，然后对频域信号求模平方”方法来实现对用户的检测。(In fact, energy detection is used for energy accumulation in a certain frequency range. If the accumulated energy is higher than a certain threshold, the existence of the signal is indicated. If the threshold is lower than the threshold, then only noise is indicated. The starting point of energy detection is that the energy of signal plus noise is greater than that of noise. The energy detection method does not make any assumptions about the signal, and it is a blind detection algorithm. This code "converts the frequency domain into the frequency domain by FFT, then calculates the square of the square of the frequency domain signal" to realize the detection of the user.)

说明：  三次样条插值函数，分别是基于二阶导数的和一阶导数实现的(Cubic spline interpolation function)

说明：  最小熵解卷积程序，适合对故障信号降噪处理，有良好的的降噪效果(Minimum entropy deconvolution program, suitable for noise reduction of fault signals)

首先将非平稳的故障振动信号进行双树复小波包分解，得 到不同频带的分量；然后对每个分量求其峭度值和相关系数并进行比较；最后选取峭度值和相关系数较大的分量 进行软阈值降噪和双树复小波包重构，即可有效地消除振动信号中噪声的干扰，同时保留信号中的有效信息即实 现了故障特征信息的提取。(In view of the above situation, a new fault diagnosis method is proposed based on dual-tree complex wavelet packet transform and threshold de-noising. Firstly, the non-stationary fault signal is decomposed into several different frequency band components through dual-tree complex wavelet packet decomposition. Secondly, Kurtosis and the cross-correlation coefficient of each component are obtained and compared. Due to the kurtosis reflecting the signal variations, if the kurtosis value is bigger, the degree of the change of signal is bigger too. The correlation coefficient can reflect the proximity between the component and the original signal at the same time, the correlation coefficient is bigger, the more similar with the original signal. Finally, the components that have a bigger value are chosen to be de-noised by a soft threshold and reconstructed by dual-tree complex wavelet packet transform. The noise interference was eliminated effectively, and the effective si)

说明：  my_noise文件是已知信号方差，加入信号噪声，可以加入高斯噪声，淑盐噪声,等其他噪声十分有用 其他两个函数时加入噪声信噪比SNR,但是有一点,这种方法并不准确,不能保证加入噪声信噪比为SNR,实际测试中可以看出,但是可以加入相同的信噪比的噪声(my_noise document is known to signal variance, by adding signal noise, Gaussian noise can be added, suk salt noise, noise, and other very useful function when the other two noise signal to noise ratio SNR, but there is one point, this method is not accurate , can not be guaranteed by adding the noise signal to noise ratio for the SNR, the actual test can be seen, but can join the same noise signal to noise ratio)

颜色预测模型评估,印刷品质量检测技术,实现光谱分析，检测色彩偏差(color model assessment, print quality inspection technology, spectrum analysis, color deviation detection)

说明：  一个EM算法实现图像分割，可以实现彩色图像分割。(An EM algorithm is used for image segmentation.)

利用八方向的sobel边缘检测，实现对图像多梯度的梯度检测，检测出的边缘信息更加平滑连续。(Eight direction Sobel edge detection is used to realize gradient detection of image multi gradient, and the edge information detected is more smooth and continuous.)

运用信息论的知识求解图像的信息量也就是熵的算法。(use of information, the knowledge to solve image is the amount of entropy algorithm.)

模糊核聚类及几篇文章，用于数据和图像的模糊聚类分割，效果还行(nuclear fuzzy clustering and articles for data and image segmentation fuzzy clustering, the results were OK)