This paper presented a new method to obtain the statistical relation between image features and matching probability. After integrating the Gabor wavelet features and some other parameters as matching area measures, t...
详细信息
This paper presented a new method to obtain the statistical relation between image features and matching probability. After integrating the Gabor wavelet features and some other parameters as matching area measures, this paper uses the support vector machine (SVM) classification method to transform the estimation of matching probability problem into a classifying one. The experiments show that the proposed method not only has faster computation speed than the method based on the correlation functions, but also gives a reasonable precise estimation.
An approach based on stroke orientation and asymmetric distribution model about feature parameter was proposed, which incorporates structural feature into statistical strategy. An improvement to the algorithm for extr...
详细信息
An approach based on stroke orientation and asymmetric distribution model about feature parameter was proposed, which incorporates structural feature into statistical strategy. An improvement to the algorithm for extracting fork points from skeleton images defends the reliability of stroke extraction. The feature vector for statistical recognition is extracted directly from stroke structure, and the asymmetric distribution model is applied to compute distances. The experimental results indicate that the proposed system is effective to handwritten Chinese character recognition.
An efficient approach of using multi-level wavelet transform and co-occurrence matrix for texture defect detection was proposed. The defective image is firstly decomposed into approximation and detail sub-images at va...
详细信息
An efficient approach of using multi-level wavelet transform and co-occurrence matrix for texture defect detection was proposed. The defective image is firstly decomposed into approximation and detail sub-images at various levels by wavelet transform. Then, the co-occurrence matrix features of the detail sub-images are computed and analyzed to decide the appropriate level at which the approximation sub-images are reconstructed into non-texture images for defect detection. The experimental results show that this approach is efficient both in speed and performance.
The purpose of trace reconstruction is to recover information about handwriting sequence from static images of characters, which helps incorporate online methods into offline applications and unify the recognition str...
详细信息
The purpose of trace reconstruction is to recover information about handwriting sequence from static images of characters, which helps incorporate online methods into offline applications and unify the recognition strategies of single character and character sequence. A stroke segment-based algorithm was proposed, which is equal to the problem of ordering the stroke segments in nature. Structural graph of stroke segments is extracted from skeleton images, by which the relational graph is created. Trace reconstruction is realized as a globally optimal problem, and the handwriting trace is considered as the path with the totally minimal orientation variance, which can be resolved by searching a Hamiltonian path with minimal cost. The cases of trace reconstruction were analyzed, and the accuracy reaches 93.5% on 200 images. The experimental results indicate the proposed approach is effective to the reconstruction of handwriting traces of handwritten numerals.
A kind of subspace clustering algorithm based on density and grids was proposed. Boundaries between classes are located by partitioning grids in the data space and finding sparse regions. The algorithm uses the search...
详细信息
A kind of subspace clustering algorithm based on density and grids was proposed. Boundaries between classes are located by partitioning grids in the data space and finding sparse regions. The algorithm uses the searching strategy of projected pursuit to find the classes in the subspace. A competitive pruning procedure is utilized to reduce the computational complexity. The experimental results on the benchmark datasets show that the proposed algorithm has advantages in accuracy and computational complexity.
We formulate single-image multi-label segmentation into regions coherent in texture and color as a MAX-SUM problem for which efficient linear programming based solvers have recently appeared. By handling more than two...
详细信息
We formulate single-image multi-label segmentation into regions coherent in texture and color as a MAX-SUM problem for which efficient linear programming based solvers have recently appeared. By handling more than two labels, we go beyond widespread binary segmentation methods, e.g., MIN-CUT or normalized cut based approaches. We show that the MAX-SUM solver is a very powerful tool for obtaining the MAP estimate of a Markov random field (MRF). We build the MRF on superpixels to speed up the segmentation while preserving color and texture. We propose new quality functions for setting the MRF, exploiting priors from small representative image seeds, provided either manually or automatically. We show that the proposed automatic segmentation method outperforms previous techniques in terms of the global consistency error evaluated on the Berkeley segmentation database.
Presents an efficient method which uses two neighboring frames in image sequences for target identification. Using statistical information about the background noise and candidate regions' noise after background r...
详细信息
ISBN:
(纸本)0780342534
Presents an efficient method which uses two neighboring frames in image sequences for target identification. Using statistical information about the background noise and candidate regions' noise after background registration, we can determine those candidate regions that have the same or similar noise distributions to the background's which should be background regions, and those candidate regions that have different noise distributions from the background's which should be the target region. In particular, when there is only one target in the image, we can simply determine that the candidate region whose noise distribution is most different from the background's is the true target.
An algorithm for unsupervised linear discriminant analysis was presented. Optimal unsupervised discriminant vectors are obtained through maximizing covariance of all samples and minimizing covariance of local k-neares...
详细信息
An algorithm for unsupervised linear discriminant analysis was presented. Optimal unsupervised discriminant vectors are obtained through maximizing covariance of all samples and minimizing covariance of local k-nearest neighbor samples. The experimental results show our algorithm is effective.
A multi-homed VPN architecture based on extended SOCKSv5 and TLS was proposed. The architecture employs a dynamic connection mechanism for multiple proxies in the end system, in which the security-demanded transmissio...
详细信息
A multi-homed VPN architecture based on extended SOCKSv5 and TLS was proposed. The architecture employs a dynamic connection mechanism for multiple proxies in the end system, in which the security-demanded transmission connections can switch smoothly among the multiple proxies by maintaining a coherent connection context. The mechanism is transparent to application programs and can support the building of VPN. With the cooperation of some other security components, the mechanism guarantees the resource availability and reliability of the end system against.some attacks to the specific ports or hosts.
This thesis presents a circuit architecture to realize clock recovery for fast Ethernet application, including system architecture, modified Mueller Muller algorithm for 100BASE-TX, phase detector for 100BASE-FX and m...
详细信息
This thesis presents a circuit architecture to realize clock recovery for fast Ethernet application, including system architecture, modified Mueller Muller algorithm for 100BASE-TX, phase detector for 100BASE-FX and multiple output charge pump PLL. The clock generator circuit with 3.3 V operation voltage has been verified by TSMC 0.35 μm 1P5M CMOS process. The results show that this clock recovery circuit can exactly extract the timing information. It has advantages over other ones for simplicity and easy implementation.
暂无评论