We consider the problem of joint motion segmentation and *** algorithm based on region-merging and the Minimum Description Length(MDL) principle is presented *** proposed technique is an unsupervised bottom-up merging...
详细信息
We consider the problem of joint motion segmentation and *** algorithm based on region-merging and the Minimum Description Length(MDL) principle is presented *** proposed technique is an unsupervised bottom-up merging method which could be separated into two parts:initial segmentation and further *** error map of motion-compensated prediction error(MCPE) of each region is used to check the motion correlation of the contiguous regions and to obtain the motion vectors of the newly merged *** MDL-based criteria are used to evaluate each possible fusion in further segmentation. The criteria are also used as a test for stopping the merging *** results demonstrate the efficiency of the proposed method.
The skeleton of an image object is a simplified representation, which is of great significance for the imagerecognition and matching. To obtain a smooth and accurate skeleton of a specified object in the gray image, ...
详细信息
The sliding window method will cause the severe unbalanced dataset problem. In this paper, under-sample the majority class method is adopted to solve this problem,and SVM is used to classify the processed data The bet...
详细信息
The sliding window method will cause the severe unbalanced dataset problem. In this paper, under-sample the majority class method is adopted to solve this problem,and SVM is used to classify the processed data The better prediction result of minority class (that is, the signal peptides positive sample set) is ***, we discover that the (-3,-1) rule is helpful to the prediction. So Information content based feature weighting method is proposed This method avoids the blindness of the previous algorithm in dealing with different sites. Experiments show that not only is the correct prediction rate of minority class improved dramatically, but also the correct prediction rate of majority class is kept in a high *** of the unbalanced data processing and the proposed information content based feature weighting method can greatly improve the performance of SVM classifier of signal peptides.
The method described here for reconstructing the shape of Lambertain surfaces from the shading information inherent in a monocular image can deal with the surface whose image,a smooth and non-self occluding one,is pro...
详细信息
ISBN:
(纸本)0780312333
The method described here for reconstructing the shape of Lambertain surfaces from the shading information inherent in a monocular image can deal with the surface whose image,a smooth and non-self occluding one,is produced by orthographic projection and parallel light source from a known *** dividing the image area into serial triangles,the intensity function I(x,y) on each triangle can be *** the help of a support function and a transformation,it is proved that each triangle represents a cylinder in a new(X-Y) coordinates with (s,t) as its components of the surface ***,the ill-posed image brightness constraint equation in cartesian coordinates can be solved in(X-Y) coordinates with only one variable(s or t).It is necessary to find out,firstly,the pixel which has the maximal intensity,the assigning of its depth leads to the direct recovery of the shape at each point without the necessity of integration.
In recent years, iris recognition has become a pro-active topic in both research and practical applications. In general, a typical iris recognition system includes iris preprocessing, feature extraction, and recogniti...
详细信息
In recent years, iris recognition has become a pro-active topic in both research and practical applications. In general, a typical iris recognition system includes iris preprocessing, feature extraction, and recognition. In this paper, an efficient iris recognition system is illustrated. The major procedure includes three steps: 1) the geometrical method is used to localize the pupil;2) in order to reduce computational cost, the outer (or limbus) boundary is localized using the shrunk image with the Hough transform and modified Canny edge detector;3) Log-Gabor filter is adopted to capture local and global characteristics of the iris. The phase of the filtered iris is recorded as features. The proposed method is implemented and tested on two iris database sets, i. e CASIA and SJTU-IDB, with different contrast quality. The experimental results show that the performance of the proposed method is encouraging.
The embedded block coding with optimized truncation (EBCOT) is the state-of-the-art coding technique for image compression, which is the heart of the latest still image compression standard JPEG2000. EBCOT can be part...
详细信息
Deep convolutional neural networks(CNNs) have recently shown impressive performance as generic representation for recognition. However, the feature extracted from global CNNs lack geometric invariance, which limits th...
详细信息
ISBN:
(纸本)9781510803084
Deep convolutional neural networks(CNNs) have recently shown impressive performance as generic representation for recognition. However, the feature extracted from global CNNs lack geometric invariance, which limits their robustness for classification and detection of highly variable *** improve the invariance of the features without degrading their discriminative power and speed up the calculation, we follow the next two method. Firstly, we adopt the scheme called multi-scale orderless pooling(MOP-CNN) which extracts CNNs activation from local patches of the image at multiple scale levels, performs orderless VLAD pooling of these activations at each level separately, and concatenates the result. Second, to speed up the calculation, we adapt the SPP-net as the CNNs architecture. Using SPP-net, we compute the feature maps from the entire image only once, and then pool features in arbitrary regions(sub-images) to generate fixed-length representations for training the detectors. This method avoids repeatedly computing the convolutional features. On the challenging SUN397 Scenes classification datasets, our method achieves competitive classification results.
Under CVL dataset, we provide an image segmentation approach based on adaptive histogram based K-means clustering and fast Hough transformation. This work firstly analyzes the characteristics of ear images in CVL face...
详细信息
A noise erosion operator based on partial differential equation (PDE) is introduced, which has an excellent ability of noise removal and edge preservation for two-dimensional (2D) gradient data. The operator is applie...
详细信息
A noise erosion operator based on partial differential equation (PDE) is introduced, which has an excellent ability of noise removal and edge preservation for two-dimensional (2D) gradient data. The operator is applied to estimate a new diffusion coefficient. Experimental results demonstrate that anisotropic diffusion based on this new erosion operator can efficiently reduce noise and sharpen object boundaries.
Iris segmentation is a critical stage in the whole iris recognition process. In this paper, a procedure of iris segmentation is presented which was designed on the basis of the natural properties of the iris. The prop...
详细信息
暂无评论