Wavelet Edge Detection Method Based-on Multi-scale Fuzzy Logic

A novel edge detection algorithm is proposed by implementing fuzzy image enhancement at different scales.The implement of this algorithm can reduce the negative influence bring from the non-maximum suppression algorithm which select only one threshold.The success of the approach depends on the selection of a general fuzzy operator(GFO) for the enhancement of contrast of region after smoothing of image at different scales.Consequently,it is more obvious that coarse intensity changes are obtained at a large scale whereas the fine details of intensity changes are obtained at a small scale.A pretty performance is obtained by used this work.     

The Signals Analysis System Based on Fast Wavelet Transform

The fast wavelet transform (FWT) algorithm in wavelet analysis was introduced in the paper. With quadrature mirror filters (QMF) associated with popular wavelet bases, the fast wavelet decomposition and reconstruction for signals were implemented. Combined with virtual instrument technique, the FWT analysis system for signals was successfully developed. The system can break up signal not only into approximations, which are the high-scale and low-frequency components of the signal, but also into details which are the low-scale and high-frequency components. Especially it can identify singularity signal, which contain some important message of equipment condition and fault, and refine signal from noisy signal, which is corrupted by noise.     

Partial discharge pattern recognition based on S transform and two-directional 2DPCA

A new feature extraction method is proposed to recognize different types of partial discharge (PD) signals. Firstly,four typical categories of PD artificial defect models are made and S transform (ST) is employed to obtain a time-frequency representation of the recorded UHF signals. Then,two-directional two-dimensional principal component analysis ((2D) 2PCA) is applied to compress the ST amplitude (STA) matrix to extract features. Finally,support vector machine (SVM) combined with particle swarm optimization (PSO) algorithm is introduced to accomplish the recognition of experimental samples. Classification results demonstrate that the average recognition rate of (10,5) combination is the highest while the one of (5,5) combination is the lowest among four kinds of feature dimension combinations. Moreover,PSO can obviously improve the classification performance of SVM. Specifically,all the average recognition rates of PSO-SVM are higher than 94.43%and the maximum value comes to 97.67%. Therefore,the feature sets extracted by ST and (2D) 2PCA can not only achieve dramatic dimension reduction,but also retain the major information of original data. It is proved that the proposed algorithm can obtain ideal results in PD pattern recognition.     

Segmentation algorithm for urinary sediment image combiningwavelet transform and 2D Maximum entropy threshold

In order to solve the problem that urine sediment visible components cannot be segmented effectively because of complex components, complicated defocusing in image and poor discrimination between object and background, a method based on combination algorithm wis designed to segment urine sediment. The wavelet transform wis used to erase the effect of defocusing. Then morphology wis utilized to get the subimages that include the particles. The segmentation method combining the wavelet transform based segmentation and the two dimensional entropy threshold based segmentation wis employed to segment urine sediment visible components. Experimental results show that the proposed method can segment urinary sediment images effectively and precisely.     

Multi scale and multi orientation features for image de noising

An adaptive algorithm for image de noising is proposed based on the multi scale and multi orientation features. The coefficients in different scales and different directions are obtained by image decomposition using the nonsubsampled contourlet transform. Then thresholds functions are adaptively set with these coefficients. The texture of the image information is introduced by using the mean of decomposition scale and the energy of regional. The greater the energy, the more information of the texture while the same decomposition scales, the smaller the threshold is set. On the contrary, the greater the threshold is set. After the de noising and then reconstruction of these coefficients, image de noising is implemented. Compare to the wavelet transform threshold and contourlet transform threshold, the nonsubsampled contourlet transform pick up the image detail better and improve the quality of the image.     

Application of Time-frequency Representations in Study on Heart Sound

Heart sound is non-stationary signal, some of its important characteristics can be obtained by time - frequency analysis method. The criteria of different time-frequency representation concentricity of spectrogram, coverage, MSE and regularity of time envelope axe proposed, the performances of Short-time Fourier transform, Wigner - Ville distribution, continuous wavelet transform, and S transform in study on heart sound are compared based on these criteria , and concrete example of application is given.     

Filtering Construct of Adaptive Filter Grou Pwith Multi-scale Decomposition

The principle and method of the adaptive filter and the filtering with wavelet transform were analyzed, and the model and method of adaptive filtering with wavelet transforms for the transient signal was established. The separated noise of signal by the multi-scale decomposition of wavelet transforms, was the input signal of adaptive filter, and accordingly the optimal filtering method of signal-noise decomposition was realized. By the adaptive filter grou Pbased on the wavelet transform, the optimal filtering to the multi-noise of signal is achieved at the same time, and the method presented in this paper has the excellent filtering capability. Examples of application demonstrate that this method presented is excellent to realize the optimal estimate to the valuable signal and noise of the transient signal in the same frequency segment.     

Identification of Acceleration Pulses of Near-Fault Earthquake Ground Motion by Wavelet

Acceleration pulses have been found in near-fault earthquake ground motions long ago and attracted lots of researchers, since such pulses, especially low frequency pulses have great impact on engineering structures. Many people previously focused on velocity pulses which result from acceleration pulses, however, recently a few researchers began to directly study acceleration pulses, such as to get high frequency pulses and low frequency pulses with empirical mode decomposition. Wavelet transform was utilized to decompose acceleration records. Subsequently high frequency and low frequency components of the records were obtained and acceleration pulses were identified. By comparisons in response spectra and structural nonlinear responses, it is proved that the decomposition of acceleration records with wavelet is reliable and the identification of acceleration pulses is correct.     

基于深度学习的多品种鲜食葡萄采摘点定位

鲜食葡萄品种多样,具有不同的形状和颜色。针对葡萄采摘机器人采摘不同品种鲜食葡萄时采摘点定位精度降低的问题,提出一种基于深度学习的多品种鲜食葡萄采摘方法。首先利用PSPNet(MobileNetv2)语义分割模型分割葡萄图像,在葡萄上方设置一个兴趣区域,在兴趣区域内使用自适应阈值果梗方向Canny边缘检测提取果梗边缘信息,然后采用霍夫变换检测果梗边缘上的直线段并进行直线拟合。最后将拟合的直线与兴趣区域的水平对称轴的交点作为采摘点。对晴天顺光、晴天逆光、晴天遮阴3种光照条件下的克瑞森、阳光玫瑰、红提和黑金手指4个品种的360幅葡萄图像进行采摘点定位试验。结果显示,采摘点定位准确率为91.94%,定位时间为187.47 ms,在模拟试验中采摘成功率为85.5%。