Papers by Keyword: Morphological Filtering

Abstract: A new method of pitch detection based on morphological filtering is proposed. Noisy speech signal is filtered by morphological filtering to remove the noise and highlight pitch, and then HHT is employed to get Hilbert-Huang spectrum and to calculate instantaneous energy and its derivative. The moment of glottal opening and closing can be accurately located through tracking mutation of instantaneous energy, so that variation of pitch period can be accurately tracked. Compared with other traditional method of pitch detection, this method not only truly describes non-stationary and non-linear characteristics of speech signal, but also it is an adaptive process for the analysis of the speech signal. The experiments showed that the method has strong anti-noise and can accurately detect the pitch of speech in low SNR.

Abstract: One rolling bearing fault diagnosis method based on minimum entropy deconvolution (MED) and morphological filtering is proposed. Firstly, the strong background noise of rolling bearings is decreased by the MED method, then the morphological filtering which have different length of structure elements is designed and applied to the de-noised signal. Subsequently, bearing’s fault characteristic frequency is extracted by the amplitude spectrum analysis to the best morphological filtering component which have the maximum kurtosis. This method is used to analyze both a simulated bearing signal and an experimental inner ring fault signal, and the good result validated the effectiveness of this method.

Abstract: The mud pulse signal extracting in MWD(Measurement while drilling) was a core technology in the oil drilling developing process. In terms of the extraction and recognition problems of weak mud pulse signal in MWD, this paper analyzed the transmission format of Manchester coded down-hole datum and transmission characteristics of mud pulse signal and used morphological filtering algorithm to extract the mud pulse signal. According to the characteristics of the Manchester coded underground synchronization signal. After determining the location of the signal start time, It proposed a pattern similar waveform-recognition algorithm to detect the mud pulse signal. The field tests show that the algorithm can accurately define the mud pulse signal starting time and can effectively improve the decoding accuracy and meet the requirements of engineering applications.

Abstract: In order to solve the problem of over-segmentation of traditional watershed algorithm, an improved watershed segmentation algorithm of the bridge image was proposed in this paper. First, the input image was filtered by top-hat transformation and bottom-hat transformation, and then, a multiscale algorithm for computing morphological gradient images is proposed, and the threshold for marker-extraction is automatically calculated according to the statistics of local extreme points in the gradient map. The watershed algorithm is applied on the modified gradient map to segment the image. Then, the over-segmentation regions of the initial watershed segmentation is settled by region merging based on fisher distance.Region merging is ended according to divergence principle. Many contrast experimental results verified the feasibility and validity of the method.

Abstract: SAR image recognition is an important content of of aviation image interpretation work. In this paper, the characteristics of SAR images a practical significance of morphological filtering neural network model and its adaptive BP learning algorithm. As can be seen through the experimental results, the algorithm can not only adapt to the complex and diverse background environment, and has a displacement of the same continuous moving target detection capability, telescopic invariant and rotation invariant features.

Abstract: For the complexity of the distribution network and the particularity of single-phase grounding fault, intelligent distribution network grounding fault line selection model based on the Extreme Learning Machine (ELM) information integration is proposed in the paper. When a single-phase grounding fault happened, the relation functions for the wavelet packet decomposition, the fifth harmonic method and the traveling wave method are respectively determined, the fault estimate data of transient zero-sequence current are calculated. The fault line is accurately judged by the ELM networks information fusion. Through analyzing MATLAB simulation result about different ground fault line selection, the validity and accuracy of the method are verified.

Abstract: In this paper, a novel fault diagnosis method for gear was approached based on morphological filter, ensemble empirical mode decomposition (EEMD), sample entropy and grey incidence. Firstly, in order to eliminate the influence of noises, the line structure element was selected for morphological filter to denoise the original signal. Secondly, denoised vibration signals were decomposed into a finite number of stationary intrinsic mode functions (IMF) and some containing the most dominant fault information were calculated the sample entropy. Finally, these sample entropies could serve as the feature vectors, the grey incidence of different gear vibration signals was calculated to identify the fault pattern and condition. Practical results show that this method can be used in gear fault diagnosis effectively.

Abstract: In this paper, a novel comprehensive fault identification approach was proposed based on the harmonic window decomposition (HWD) frequency band energy extraction and grey relation degree. Firstly, in order to eliminate the influence of noises, the line structure element was selected for morphological filter to denoise the original signal. Secondly, due to the energy of vibration signal will change in different frequency bands when fault occurs, therefore, the six feature frequency bands which contain the typical fault information were extracted by harmonic window decomposition that need not decomposition; and the energy distribution of each band could be calculated. Finally, these energy distributions could serve as the feature vectors, the grey relation degree of different vibration signals was calculated to identify the fault pattern and condition. Practical results show that this method can identify rotor fault patterns effectively.

Abstract: A new method of speech enhancement is proposed based on morphological filter and wavelet transform. The system begins by first conducting morphological filtering, then distinguishing the unvoiced, voiced and noise using TEO in the wavelet domain. It then executes wavelet transform using different threshold on multiscale, and at the same time to improve the threshold function. Experimental results showed that the method not only suppressed noise effectively but also reduced the loss of the unvoiced. It also not only enhanced SNR, but also improved voice clarity and comfort. The merits it espouses makes it an effective speech enhancement algorithm.

Abstract: This paper has studied infrared target background suppression technology in the case of low SNR. In order to meet the high detection rate and the low false alarm rate, this paper presents a new infrared image background prediction method based on morphological anisotropic diffusion filtering. Experiments show that the proposed algorithm, under the condition of rolling background infrared image, has a good background predication and targeting enhancement.