Search

To better capture the local geometric structural information of 3D point clouds, an adaptive graph convolutional 3D point cloud recognition algorithm is proposed based on attention mechanism. To address the drawback of fixed convolutional kernels ignoring features, the algorithm first dynamically learns adaptive convolutional kernels based on graph structural features. Furthermore, to enhance the modeling capability of the model for local geometric structures, the weight distribution of the convolutional kernels using a vector attention mechanism is adjusted adaptively. Subsequently, a graph is constructed using the position features of the point cloud and perform convolution operations on the newly constructed graph structural features using the adaptive convolutional kernels. Finally, new point cloud features through pooling is obtained. Experimental results demonstrate that the proposed algorithm effectively extracts local geometric structural information and achieves higher accuracy in classification tasks even with a limited number of sampled points, outperforming previous point cloud convolutional algorithms. The proposed algorithm also exhibits certain advantages compared to existing methods for point cloud classification and segmentation, as evidenced by the performance evaluation on the ModelNet40, ScanObjectNN, and ShapeNetPart datasets. [ABSTRACT FROM AUTHOR]

In order to improve the DOA estimation accuracy of mixed signals and reduce array aperture loss, a high-precision DOA estimation algorithm based on oblique projection operator is proposed. The proposed algorithm estimates the independent and coherent signals of the mixed signals in two stages. Firstly, the covariance matrix of the data received by the array element is processed by the estimating signal parameter via rotational invariance techniques(ESPRIT), and the DOA estimates of the independent signals in the mixed signals are calculated. Then, the algorithm uses the oblique projection operator to remove the independent signal information in the mixed signals to obtain a new covariance matrix. The signal subspace of the newly obtained covariance matrix is used for decoherence processing. Finally, the ESPRIT algorithm is used to calculate the DOA estimates of the coherent signals. Simulation results show that the proposed algorithm has higher accuracy than traditional mixed signal DOA estimation algorithms in the case of low signal-to-noise ratio and small signal incidence interval, and effectively reduces array aperture loss. Under the condition of different sampling snapshot numbers, the proposed algorithm also shows stronger robustness. [ABSTRACT FROM AUTHOR]

Aiming at the precision loss problem of most low-complexity Root-MUSIC algorithms at present, a low-complexity Root-MUSIC algorithm with precision compensation ability is studied and proposed. The algorithm reconstructs the autocorrelation matrix with Toeplitz shape according to the first row of the approximate data observation matrix obtained by finite snapshots, so that the reconstructed autocorrelation matrix has Hermitian property. After decomposing the reconstructed autocorrelation matrix, the noise subspace is obtained, the noise subspace is flipped and split, a new root-finding polynomial is reconstructed, and then the DOA estimated value is obtained by the root-finding method. The algorithm proposed in this paper using Toeplitz matrix reconstruction and root polynomial reduction effectively improves the DOA estimation accuracy of the improved Root-MUSIC algorithm. And the time complexity of the improved algorithm is no higher than that of previous algorithms. Under different incident sources and sampling snapshots, the algorithm proposed in this paper also shows stronger robustness and stability. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Northeastern University (Natural Science) is the property of Dongbei Daxue Xuebao and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Northeastern University (Natural Science) is the property of Dongbei Daxue Xuebao and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Northeastern University (Natural Science) is the property of Dongbei Daxue Xuebao and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Tｈｉｓ ｐａｐｅｒ ｐｒｏｐｏｓｅｓ ａｎ ａｌｇｏｒｉｔｈｍ ｂａｓｅｄ ｏｎ ｏｒｉｅｎｔａｔｉｏｎ ｓｃｏｒｅｓ ｃｏｍｂｉｎｅｄ ｗｉｔｈ Ｆｒａｎｇｉ ｆｉｌｔｅｒ ｔｏ ｆｉｘ ｔｈｅ ｄｉｆｆｉｃｕｌｔｉｅｓ ｃａｕｓｅｄ ｂｙ ｂｉｆｕｒｃａｔｉｏｎ ａｎｄ ｃｒｏｓｓｉｎｇｓ ｏｆ ｒｅｔｉｎａｌ ｖｅｓｓｅｌｓ. Ｔｈｅ Ｆｒａｎｇｉ ｆｉｌｔｅｒꎬ ｗｈｉｃｈ ｉｓ ｃｏｎｓｔｒｕｃｔｅｄ ｂｙ Ｈｅｓｓｉａｎ ｍａｔｒｉｘ ａｎｄ ｓｕｉｔａｂｌｅ ｆｏｒ ｆｉｌｔｅｒｉｎｇ ｔｈｅ ｌｉｎｅａｒ ｏｂｊｅｃｔꎬ ｉｓ ｕｓｅｄ ｔｏ ｅｎｈａｎｃｅ ｔｈｅ ｃｏｎｔｒａｓｔ ｏｆ ｔｈｅ ｂｌｏｏｄ ｖｅｓｓｅｌｓ. Ｔｈｅ ａｎｉｓｏｔｒｏｐｙ ａｎｄ ｏｒｔｈｏｇｏｎａｌｉｔｙ ｏｆ ｔｈｅ ｃａｋｅ ｗａｖｅｌｅｔ ｉｎ ｔｈｅ ｏｒｉｅｎｔａｔｉｏｎ ｓｃｏｒｅｓ ａｒｅ ｕｓｅｄ ｔｏ ｆｉｌｔｅｒ ｔｈｅ ｂｌｏｏｄ ｖｅｓｓｅｌｓ ｉｎ ｍｕｌｔｉｐｌｅ ａｎｇｌｅｓ ａｎｄ ｄｉｒｅｃｔｉｏｎｓꎬ ｗｈｉｃｈ ｉｓ ｂｅｎｅｆｉｃｉａｌ ｔｏ ｔｈｅ ｐｒｏｃｅｓｓｉｎｇ ｏｆ ｄｅｔａｉｌｓ ａｎｄ ｔｈｅ ｃｏｍｐｌｅｔｅ ｓｅｇｍｅｎｔａｔｉｏｎ ｏｆ ｔｈｅ ｒｅｔｉｎａｌ ｖａｓｃｕｌａｒ ｎｅｔｗｏｒｋ. Ｔｈｅ ｃｏｍｐａｒｉｓｏｎ ｏｆ ｔｈｅ ｍｅｔｈｏｄ ｐｒｏｐｏｓｅｄ ｗｉｔｈ ｏｔｈｅｒ ｅｘｉｓｔｉｎｇ ｍｅｔｈｏｄｓ ｓｈｏｗｓ ｔｈａｔ ｔｈｅ ａｌｇｏｒｉｔｈｍ ｉｓ ｂｅｔｔｅｒ ｔｈａｎ ｏｔｈｅｒ ａｌｇｏｒｉｔｈｍｓ ｉｎ ｐｒｏｃｅｓｓｉｎｇ ｔｈｅ ｂｉｆｕｒｃａｔｉｏｎ ａｎｄ ｃｒｏｓｓｉｎｇｓꎬ ａｎｄ ｔｈｅ ｍｅａｓｕｒｅｍｅｎｔｓ ｏｆ ａｃｃｕｒａｃｙꎬ ｓｅｎｓｉｔｉｖｉｔｙ ａｎｄ ｓｐｅｃｉｆｉｃｉｔｙ ａｒｅ ｓｕｐｅｒｉｏｒ ｔｏ ｅｘｉｓｔｉｎｇ ａｌｇｏｒｉｔｈｍｓ. [ABSTRACT FROM AUTHOR]

A method based on multifractal detrended fluctuation analysis(DFA)， which is adaptable for unsteady images， is introduced to segment retina images. In this method， histogram equalization is first used to precondition images for the enhancement of blood vessels， and then， the generalized Hurst index of images is calculated by multifractal DFA. The blood vessel index is used to segment the blood vessels. Finally， morphological postprocessing is used to get the final blood vessel images. Experiments based on DIARETDE0 and DIARETDE1 databases show that the proposed method has better integrity and connectivity in the treatment of retinal lesions， and it can extract the main body of blood vessels better， which has a good clinical value. [ABSTRACT FROM AUTHOR]

To improve the imaging quality in the minimum variance method of lowcomplexity， an improved low-complexity amplitude and phase estimation beamforming algorithm was proposed. The adaptive weighted value was calculated by extracting the effective rows of the covariance matrix， where the selection of the effective number of rows was determined by a more accurate Gaussian phase coherence coefficient， which was then used to modify the weighting vector in order to reduce the complexity and ensure the quality of the imaging at the same time. Through experimental verification， the improved algorithm was compared with the original algorithm and another beamspace method to reduce the complexity， which fully proves the superiority of the improved algorithm in reducing the complexity and improving the imaging quality. [ABSTRACT FROM AUTHOR]

In medical ultrasound imaging， amplitude and phase estimation (APES) beamforming algorithm makes the amplitude estimation of the desired signal more robustly， but it is at the expense of resolution. To solve this problem， the adaptive beamforming algorithm was proposed which was based on APES and was combined with subspace projection. Firstly， the weighted vector was calculated by the amplitude phase estimation algorithm， and then the weighted vector was projected to the desired signal subspace of the echo signal to suppress the noise and interference， and to improve the resolution and contrast of the image. The simulation results show that the proposed algorithm is superior to the original amplitude and phase estimation algorithm in terms of resolution， contrast and robustness. [ABSTRACT FROM AUTHOR]

Copyright of Research & Exploration in Laboratory is the property of Research & Exploration in Laboratory and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Compressed sensing (CS) is of great significance to solving such problems as high sampling rate, huge storage pressure and long processing time in the process of stepped-frequency continuous wave ground penetrating radar (SFCW-GPR). Aiming at the problem that block objects can’t meet the sparseness in the detecting area, and using the orthogonal basis for sparse processing of block objects to satisfy the sparsity condition, a new observation matrix that was suitable for block objects was formed by combining the dictionary matrix and sparse matrix. The sparse coefficients were solved by using the compressed sensing convex optimization algorithm. Finally, the reflection coefficients of block objects were obtained through sparse transformation of the sparse coefficients. The simulation results showed that the proposed method is feasible and has higher accuracy and resolution ratio compared with the compressed sensing reconstruction model without sparsity. [ABSTRACT FROM AUTHOR]

In retinal vascular images, characteristic of image, such as low contrast, will interfere with the normal image segmentation. In allusion to the image characteristics of retinal blood vessels, an enhancement algorithms of retinal blood vessels was proposed based on improved curvelet conversion. Firstly, a preprocessing was performed with the green channel of retinal image, combined with the adaptive enhancement of transform coefficient to achieve the process. The improved morphological transformation was also used to suppress background and enhance the vessel image information with more brightness. With the comparison to other methods, the experiment result showed the validity and the values of the method in improving contrast and reducing noise. [ABSTRACT FROM AUTHOR]