Applied Mechanics and Materials Vols. 303-306

Abstract: In this paper, we studied voice signal with Gaussian noise reduction. Based on signal analysis and reconstruction principle. Application of Hilbert-Huang transform(HHT), analyzed voice signal with Gaussian noise. And comprised with wavelet transform(WT), obtained the correlation coefficient between HHT noise reduction signal and excluding the noise signal was 0.8986, WT was 0.7889. The results showed that in the voice signal with Gaussian noise, compared HHT and wavelet analysis correlation analysis values, HHT noise reduction ability was 10% higher than WT. This paper provided a new analytic method to the voice signal noise reduction and enhanced the accuracy of it.

Abstract: Spectral measurement is a common method for color measurement. Spectral data of color samples gathered by spectrometers contains a series of noises. The traditional de-noising methods have their limitation in dealing with such signal. In this paper, a de-noising method is proposed for spectral signal based on Wavelet Transform. Compared the performance of our de-nosing method with the traditional method, the results show that our method a better effect.

Abstract: Automatic and quickly extraction of bridge information from LiDAR data is of great significance in building 3D digital city and virtual earth. Especially the extraction of bridge outline is a crucial problem. It is a concern of many scholars research focus. This paper presented a method of bridge extraction using airborne LiDAR data. The biggest advantage of the method is based on priori-knowledge and by analyzing the spatial structural characteristics and geometric characteristics of the bridge. Experiments show that this method has a good accuracy compared with the result of expert interpretation.

Abstract: The image stitching method is widely used into the suspect's footprint information extraction. In order to improve the image detail and the matching precision, the Footprint map image stitching method which is based on the wavelet transform and the SIFT feature matching is put forward. The wavelet transform in this method is perform based on the pretreatment of image, move the low frequency wavelet coefficient to zero, adjusting thresholds of the high frequency wavelet coefficient and inverse transformation, then, use the SIFT to extract and match the key-points of the processed images. For the error matching pair of coarse match, you can use the RANSAC to filter them out. This article demonstrates its advantage through to the original image splicing comparisons. The experimental results show that the method display more clear detail and the precision of matching than the original method.

Abstract: Object detection is quite an important research in remote sensing image analysis. In this paper, we propose an edge and region based model for high resolution remote sensing image segmentation with level set formulation. Our method firstly made an image enhancement based on ROI (Region of Interest). By introducing the edge speed-up function, we can save time through decreasing the iterations and get a flexible segmentation considering the complexity of high resolution remote sensing image. Our method has been preliminarily applied to QuickBird and aerial images.

Abstract: A iris recognition system based on DSP is designed in this paper. A fast algorithm designed in this system for iris feature recognition. First, a point is given within the pupil randomly, and then calculate the radius from the edge detection. So the edge of pupil is drawn out. The same method is applied to find the edges of the iris. After fitting the edge, a cylindrical is drawn. The pretreatment of normalization and feature extraction is carried out using this arithmetic. This algorithm is simulated in Matlab. The results show that this algorithm is effective. In the end through program, this algorithm is run in the DSP. The run result shows it is efficient and stable.

Abstract: Comprehensive analysis of the tower in the image feature information with Harris corner imprecise for complex image, we propose an improved algorithm, which use of the improved Harris detection method to calculate the number of gray similar to the target pixel 8 field, analyze the gray level distribution of pixels within the local range, and then use the lag suppression method to set threshold , if corner is the objective function greater than the set threshold, the change point was identified as the final corner. The experimental results proved to be effective to determine the image of the tower area.

Abstract: This paper proposes a new adaptive filtering algorithm based on the p-TA-QR-LS algorithm [1]. With a coefficient-derivative-based switching scheme, the new algorithm can work between two modes (p=1 and N) and achieve overall optimum convergence performance. The resultant switching p-TA-QR-LS algorithm is thus particularly suitable for acoustic echo cancellation (AEC) where both fast convergence rate and low steady-state estimate error are desired. Experiments are conducted to verify its improved overall convergence performance.

Abstract: To effectively recognize gait signal between healthy people and patients with Parkinson, a gait signal recognition model is established based on neural network of error back propagation (EBP), and a method is proposed to effectively extract characteristic parameters. In this paper, coefficient of variation is applied in the research of gait-pressure multi-characteristic parameters through gait-pressure signal, and the neural network model can automatically recognize gait-pressure characteristics between healthy people and patients with Parkinson. This can contribute to the recognition and diagnosis of patients with Parkinson. Experiment results show a recognition rate of 90%.

Abstract: Proposed a new algebra interpolation polynomial with preferable stability, analyzed the related properties as well as stability and computational complexity, etc. Proved the new algebra interpolation polynomial can approximate any continuous functions, and it can be used to calculate the high order derivative without Runge phenomenon.