Local Invariant Features Research Articles

Fast contour torque features based recognition in laser active imaging system

Recent years laser active imaging system are widely used as a useful detection artifice. While many denoising and enhancing algorithms have been proposed, target recognition in laser active imaging is still a new domain that few researches have been done. Classical recognition methods often break done due to the characteristic of laser active imaging. In this paper we present a novel target recognition method based on fast contour torque features (FCTF). The proposed fast contour torque features contain abundant information of the target, such as the size, position, darkness and shape regularly of the contours. Meanwhile the features are invariant to rotation, scaling and affine transform, and can be computed efficiently.We first extract feature regions by MSER algorithm, and transform them into circular areas, so the regions are invariant to affine transform. Then local invariant features of the feature regions were extracted by fast contour torque feature descriptor and we input into the trained SVM classifier for identify. Comprehensive experiments show that our approach achieves higher recognition rate in rotation and affine transformation as state-of-art method, and meet the real-time requirement in laser active imaging.

Exposing region duplication through local geometrical color invariant features

Many advanced image-processing softwares are available for tampering images. How to determine the authenticity of an image has become an urgent problem. Copy–move is one of the most common image forgery operations. Many methods have been proposed for copy–move forgery detection (CMFD). However, most of these methods are designed for grayscale images without any color information used. They are usually not suitable when the duplicated regions have little structure or have undergone various transforms. We propose a CMFD method using local geometrical color invariant features to detect duplicated regions. The method starts by calculating the color gradient of the inspected image. Then, we directly take the color gradient as the input for scale invariant features transform (SIFT) to extract color-SIFT descriptors. Finally, keypoints are matched and clustered before their geometrical relationship is estimated to expose the duplicated regions. We evaluate the detection performance and computational complexity of the proposed method together with several popular CMFD methods on a public database. Experimental results demonstrate the efficacy of the proposed method in detecting duplicated regions with various transforms and poor structure.

Traffic anomaly detection based on image descriptor in videos

The huge and ever growing volume of traffic video poses a compelling demand for efficient automatic detection of traffic anomaly. In this paper, a new traffic anomaly detection algorithm is introduced. It firstly divides a traffic video into several video cubes in temporal domain, and each video cube is divided into video blocks in spatial domain. Each image block of a video block is described using the local invariant features and the visual codebook approach. Based on the descriptor of the image block, we count the category number of the block (CNB) of a video block. Then, a Gaussian distribution model for estimating the probability of normal traffic with respect to the CNB is learned. The learned Gaussian distribution model is then used to detect the traffic anomaly from the test traffic video. Eventually, the results of all video blocks are fused to achieve the final decision. Experimental results show that the proposed algorithm performs better than two existing algorithms on both the intersection traffic videos and main road traffic videos.

A new robust digital watermarking using local polar harmonic transform

Geometric distortions are more difficult to tackle than other types of attacks. It is a challenging work to design a robust image watermarking scheme against geometric distortions. In this paper, we propose a robust digital image watermarking scheme based on local polar harmonic transform. The proposed scheme has the following advantages: (1) the stable and uniform image feature points are extracted by the improved speeded-up robust feature (SURF) detector, in which the probability density gradient is utilized, (2) the affine invariant local feature regions are constructed adaptively according to the variation of local probability density, and (3) a new and effective 2D transform, named polar harmonic transform (PHT), is introduced to embed watermark in the digital image. Experiments are carried out on a digital image set of 100 images collected from Internet, and the preliminary results show that the proposed image watermarking is not only invisible and robust against common image processing operations such as filtering, noise adding, and JPEG compression, but also robust against the geometric distortions.

Fast representation based on a double orientation histogram for local image descriptors.

In recent years, extensive research on local invariant features has been conducted, and many novel descriptors have been developed for different scenarios. Frequently, these descriptors can offer unique advantages (such as stability, precision, and speed) for select applications but perform unsatisfactorily from a comprehensive perspective. Consequently, a novel local image descriptor, named fast representation using a double orientation histogram (FRDOH), is developed in this paper based on existing descriptors. First, a region is divided using intensity order (as in the local intensity order pattern descriptor) to encode spatial information. Then, the discriminability of the descriptor is enhanced using our proposed double orientation histogram. Second, to further improve the discriminability of the descriptor, the Hellinger distance is used to balance the effects of large and small bins in the histogram for the similarity measure of features. Finally, a novel interpolation strategy known as rapidly cascaded interpolation is used to calculate the intensity of the neighboring points to reduce the computation time, while achieving high precision. The performance of the developed descriptor is evaluated via numerous experiments on the affine covariant feature data set of the Oxford data set, a subset of a 3D object data set, and a subset of the IIT Delhi Touchless Palmprint data set. These experiments demonstrate that the developed FRDOH descriptor outperforms the state-of-the-art descriptors in terms of comprehensive performance.

An invariant interest point detector under image affine transformation

For vision-based mobile robot navigation, images of the same scene may undergo a general affine transformation in the case of significant viewpoint changes. So, a novel method for detecting affine invariant interest points is proposed to obtain the invariant local features, which is coined polynomial local orientation tensor (PLOT). The new detector is based on image local orientation tensor that is constructed from the polynomial expansion of image signal. Firstly, the properties of local orientation tensor of PLOT are analyzed, and a suitable tuning parameter of local orientation tensor is chosen so as to extract invariant features. The initial interest points are detected by local maxima search for the smaller eigenvalues of the orientation tensor. Then, an iterative procedure is used to allow the initial interest points to converge to affine invariant interest points and regions. The performances of this detector are evaluated on the repeatability criteria and recall versus 1-precision graphs, and then are compared with other existing approaches. Experimental results for PLOT show strong performance under affine transformation in the real-world conditions.

Rotation and scale invariant shape context registration for remote sensing images with background variations

Multitemporal remote sensing images generally suffer from background variations, which significantly disrupt traditional region feature and descriptor abstracts, especially between pre and postdisasters, making registration by local features unreliable. Because shapes hold relatively stable information, a rotation and scale invariant shape context based on multiscale edge features is proposed. A multiscale morphological operator is adapted to detect edges of shapes, and an equivalent difference of Gaussian scale space is built to detect local scale invariant feature points along the detected edges. Then, a rotation invariant shape context with improved distance discrimination serves as a feature descriptor. For a distance shape context, a self-adaptive threshold (SAT) distance division coordinate system is proposed, which improves the discriminative property of the feature descriptor in mid-long pixel distances from the central point while maintaining it in shorter ones. To achieve rotation invariance, the magnitude of Fourier transform in one-dimension is applied to calculate angle shape context. Finally, the residual error is evaluated after obtaining thin-plate spline transformation between reference and sensed images. Experimental results demonstrate the robustness, efficiency, and accuracy of this automatic algorithm.

Image stitching using chaos‐inspired dissimilarity measure

A novel image-stitching algorithm is proposed that uses the chaos-inspired dissimilarity measure. Previous approaches to image stitching generally employed invariant local features and random sample consensus (RANSAC), to verify image matches. In panoramic stitching, however, RANSAC is highly sensitive to illumination difference between the pair of images stemming from two different exposures, which essentially results in mismatched pairs, leading to poor stitching performance. By applying chaos theory, the contextual change of an image becomes chaos-like, and results in a complex fractal trajectory in phase space. The specific chaos behaviour between a pair of images is exploited, whereby the resulting phase space exhibits strong invariance to illumination changes. Accordingly, the chaos-inspired dissimilarity between images is measured, thereby the matched pairs are robustly found under the presence of illumination difference. Representative performance experiments demonstrate significant improvements over the conventional method.

基于局部不变特征的目标自动识别

In order to recognize targets in images fast and truly,an automatic target recognition method was proposed based on image entropy and speed up robust feature.First,image entropy was computed in different blocks,and regions full of texture were filtered out by threshold.The local key points in regions of interest were extracted by incorporating the Hessian and Harris detectors.Then,feature descriptors were established and principle component analysis was employed to reduce the dimensionality.Finally,nearest neighbor distance ratio classifier was explored in double directions and wrong matches were eliminated by random sample consensus.The experiment results demonstrate that the recognition rates for images in simulation database with varied view-points,scales and illuminations are 87.12%,75.31%and 84.98%,and the computing time is 70.35 ms,71.27 ms and 220.63 ms,respectively.Moreover,the correct matching rate for an aerial large planar array image of 8 956×6 708 pixels is 78.13% and the computing time is 68.09 s.Compared with speed up robust feature,the proposed method performs better both in recognition rates and computing time.

激光主动成像目标识别

Due to the characteristic of images in laser active imaging,a novel target recognition method based on fast contour torque features( FCTF) is proposed. The concept of torque is introduced into target recognition. The proposed fast contour torque features contain abundant information such as the size,position,shape regularly of the contours and darkness of the target,which are as well invariant to rotation and scaling. Meanwhile the fast calculation method greatly improves the computational efficiency. Firstly feature regions are detected using Maximally Stable Extremal Regions( MSER) algorithm,and transformed into circular areas. Then local invariant features of the feature regions are extracted by fast contour torque feature descriptor. At last thefeatures are input into the trained Suppor Vector Machine( SVM) classifier for identification. The experimental results indicate that compared with the existing laser active imaging recognition algorithms,the proposed method acquires higher recognition rate in rotation and affine transformation,and the average computing time of single frame is 9. 68 ms,which meet the real-time requirement in laser active imaging.

Increase Efficiency of SURF using RGB Color Space

SURF is one of the most robust local invariant feature descriptors. SURF is implemented mainly for gray images. However, color presents important information in the object description and matching tasks as it clearly in the human vision system. Many objects can be unmatched if their color contents are ignored. To overcome this drawback this paper proposed a method CSURF (Color SURF) that combines features of Red, Green and blue layers to detect color objects. It edits matched process of SURF to be more efficient with color space. Experimental results show that CSURF is more precious than traditional SURF and CSURF invariant to RGB color space

Region-Based Object Recognition by Color Segmentation Using a Simplified PCNN.

In this paper, we propose a region-based object recognition (RBOR) method to identify objects from complex real-world scenes. First, the proposed method performs color image segmentation by a simplified pulse-coupled neural network (SPCNN) for the object model image and test image, and then conducts a region-based matching between them. Hence, we name it as RBOR with SPCNN (SPCNN-RBOR). Hereinto, the values of SPCNN parameters are automatically set by our previously proposed method in terms of each object model. In order to reduce various light intensity effects and take advantage of SPCNN high resolution on low intensities for achieving optimized color segmentation, a transformation integrating normalized Red Green Blue (RGB) with opponent color spaces is introduced. A novel image segmentation strategy is suggested to group the pixels firing synchronously throughout all the transformed channels of an image. Based on the segmentation results, a series of adaptive thresholds, which is adjustable according to the specific object model is employed to remove outlier region blobs, form potential clusters, and refine the clusters in test images. The proposed SPCNN-RBOR method overcomes the drawback of feature-based methods that inevitably includes background information into local invariant feature descriptors when keypoints locate near object boundaries. A large number of experiments have proved that the proposed SPCNN-RBOR method is robust for diverse complex variations, even under partial occlusion and highly cluttered environments. In addition, the SPCNN-RBOR method works well in not only identifying textured objects, but also in less-textured ones, which significantly outperforms the current feature-based methods.

Object Tracking by Oversampling Local Features.

In this paper, we present the ALIEN tracking method that exploits oversampling of local invariant representations to build a robust object/context discriminative classifier. To this end, we use multiple instances of scale invariant local features weakly aligned along the object template. This allows taking into account the 3D shape deviations from planarity and their interactions with shadows, occlusions, and sensor quantization for which no invariant representations can be defined. A non-parametric learning algorithm based on the transitive matching property discriminates the object from the context and prevents improper object template updating during occlusion. We show that our learning rule has asymptotic stability under mild conditions and confirms the drift-free capability of the method in long-term tracking. A real-time implementation of the ALIEN tracker has been evaluated in comparison with the state-of-the-art tracking systems on an extensive set of publicly available video sequences that represent most of the critical conditions occurring in real tracking environments. We have reported superior or equal performance in most of the cases and verified tracking with no drift in very long video sequences.

Tree-based Morse regions: a topological approach to local feature detection.

This paper introduces a topological approach to local invariant feature detection motivated by Morse theory. We use the critical points of the graph of the intensity image, revealing directly the topology information as initial interest points. Critical points are selected from what we call a tree-based shape-space. In particular, they are selected from both the connected components of the upper level sets of the image (the Max-tree) and those of the lower level sets (the Min-tree). They correspond to specific nodes on those two trees: 1) to the leaves (extrema) and 2) to the nodes having bifurcation (saddle points). We then associate to each critical point the largest region that contains it and is topologically equivalent in its tree. We call such largest regions the tree-based Morse regions (TBMRs). The TBMR can be seen as a variant of maximally stable extremal region (MSER), which are contrasted regions. Contrarily to MSER, TBMR relies only on topological information and thus fully inherit the invariance properties of the space of shapes (e.g., invariance to affine contrast changes and covariance to continuous transformations). In particular, TBMR extracts the regions independently of the contrast, which makes it truly contrast invariant. Furthermore, it is quasi-parameter free. TBMR extraction is fast, having the same complexity as MSER. Experimentally, TBMR achieves a repeatability on par with state-of-the-art methods, but obtains a significantly higher number of features. Both the accuracy and robustness of TBMR are demonstrated by applications to image registration and 3D reconstruction.

Learning Salient Features for Speech Emotion <newline/>Recognition Using Convolutional <newline/>Neural Networks

As an essential way of human emotional behavior understanding, speech emotion recognition (SER) has attracted a great deal of attention in human-centered signal processing. Accuracy in SER heavily depends on finding good affect- related , discriminative features. In this paper, we propose to learn affect-salient features for SER using convolutional neural networks (CNN). The training of CNN involves two stages. In the first stage, unlabeled samples are used to learn local invariant features (LIF) using a variant of sparse auto-encoder (SAE) with reconstruction penalization. In the second step, LIF is used as the input to a feature extractor, salient discriminative feature analysis (SDFA), to learn affect-salient, discriminative features using a novel objective function that encourages feature saliency, orthogonality, and discrimination for SER. Our experimental results on benchmark datasets show that our approach leads to stable and robust recognition performance in complex scenes (e.g., with speaker and language variation, and environment distortion) and outperforms several well-established SER features.

Robust multimodal 2D and 3D face authentication using local feature fusion

In this work, we present a robust face authentication approach merging multiple descriptors and exploiting both 3D and 2D information. First, we correct the heads rotation in 3D by iterative closest point algorithm, followed by an efficient preprocessing phase. Then, we extract different features namely: multi-scale local binary patterns (MSLBP), novel statistical local features (SLF), Gabor wavelets, and scale invariant feature transform (SIFT). The principal component analysis followed by enhanced fisher linear discriminant model is used for dimensionality reduction and classification. Finally, fusion at the score level is carried out using two-class support vector machines. Extensive experiments are conducted on the CASIA 3D faces database. The evaluation of individual descriptors clearly showed the superiority of the proposed SLF features. In addition, applying the (\(\hbox {3D} + \hbox {2D}\)) multimodal score level fusion, the best result is obtained by combining the SLF with the \(\hbox {MSLBP}+\hbox {SIFT}\) descriptor yielding in an equal error rate of 0.98 % and a recognition rate of \(\hbox {RR} = 97.22\,\%\).

CSIFT based locality-constrained linear coding for image classification

In the past decade, SIFT descriptor has been witnessed as one of the most robust local invariant feature descriptors and widely used in various vision tasks. Most traditional image-classification systems depend on the gray-based SIFT descriptors, which only analyze the gray level variations of the images. Misclassification may happen since their color contents are ignored. In this article, we concentrate on improving the performance of existing image-classification algorithms by adding color information. To achieve this purpose, different kinds of colored SIFT descriptors are introduced and implemented. locality-constrained linear coding (LLC), a state-of-the-art sparse coding technology, is employed to construct the image-classification system for the evaluation. Moreover, we propose a simple $$\ell _2$$ l 2 -norm regularized local distance to improve the traditional LLC method. The real experiments are carried out on several benchmarks. With the enhancements to color SIFT and $$\ell _2$$ l 2 -norm regularization, the proposed image-classification system obtains approximately $$2\,\%$$ 2 % improvement of classification accuracy on the Caltech-101 dataset and approximately $$5\,\%$$ 5 % improvement of classification accuracy on the Caltech-256 dataset.

Palm vein recognition based on multi-sampling and feature-level fusion

For contactless palm vein images, particularly substantial changes in hand positioning between one image and the next, it is difficult to achieve a satisfactory recognition performance with geometry- or statistics-based methods due to issues such as non-uniform illumination and affine transformation. Therefore, with a multi-sampling and feature-level fusion strategy, a novel palm vein recognition method based on local invariant features is presented for addressing the aforementioned issues. As most of the contactless palm vein images are unclear and have low contrast, if a SIFT algorithm is directly adopted for feature extraction on the center region of a palm vein image, it will be difficult to obtain sufficient features for effective recognition. Therefore, in this paper, we first propose to take the entire palm as a Region of Interest (ROI) and perform new hierarchical enhancement on the ROI to ensure that additional features will be obtained from the subsequent feature extraction. Then, we take full advantage of the multiple samples collected in the registration stage to generate the registered template using feature-level fusion. Finally, bidirectional matching is proposed for mismatch removal. The experiments on the CASIA Palm vein Image Database and our palm vein database collected under the posture changes show that the proposed method was superior in terms of recognition performance, especially for palm vein image recognition under remarkable posture changes. In particular, the values of Equal Error Rate (EER) on the aforementioned two databases were 0.16% and 0.73%, respectively.

Compounding local invariant features and global deformable geometry for medical image registration.

Using deformable models to register medical images can result in problems of initialization of deformable models and robustness and accuracy of matching of inter-subject anatomical variability. To tackle these problems, a novel model is proposed in this paper by compounding local invariant features and global deformable geometry. This model has four steps. First, a set of highly-repeatable and highly-robust local invariant features, called Key Features Model (KFM), are extracted by an effective matching strategy. Second, local features can be matched more accurately through the KFM for the purpose of initializing a global deformable model. Third, the positional relationship between the KFM and the global deformable model can be used to precisely pinpoint all landmarks after initialization. And fourth, the final pose of the global deformable model is determined by an iterative process with a lower time cost. Through the practical experiments, the paper finds three important conclusions. First, it proves that the KFM can detect the matching feature points well. Second, the precision of landmark locations adjusted by the modeled relationship between KFM and global deformable model is greatly improved. Third, regarding the fitting accuracy and efficiency, by observation from the practical experiments, it is found that the proposed method can improve % of the fitting accuracy and reduce around 50% of the computational time compared with state-of-the-art methods.

Contact-free palm-vein recognition based on local invariant features.

Contact-free palm-vein recognition is one of the most challenging and promising areas in hand biometrics. In view of the existing problems in contact-free palm-vein imaging, including projection transformation, uneven illumination and difficulty in extracting exact ROIs, this paper presents a novel recognition approach for contact-free palm-vein recognition that performs feature extraction and matching on all vein textures distributed over the palm surface, including finger veins and palm veins, to minimize the loss of feature information. First, a hierarchical enhancement algorithm, which combines a DOG filter and histogram equalization, is adopted to alleviate uneven illumination and to highlight vein textures. Second, RootSIFT, a more stable local invariant feature extraction method in comparison to SIFT, is adopted to overcome the projection transformation in contact-free mode. Subsequently, a novel hierarchical mismatching removal algorithm based on neighborhood searching and LBP histograms is adopted to improve the accuracy of feature matching. Finally, we rigorously evaluated the proposed approach using two different databases and obtained 0.996% and 3.112% Equal Error Rates (EERs), respectively, which demonstrate the effectiveness of the proposed approach.