Performance study of feature descriptors for human detection on depth map

Depth map contains the space information of objects and is almost free from the influence of light,and it attracts many research interests in the field of machine vision used for human detection.Therefore,hunting a suitable image feature for human detection on depth map is rather attractive.In this paper,we evaluate the performance of the typical features on depth map.A depth map dataset containing various indoor scenes with human is constructed by using Microsoft’s Kinect camera as a quantitative benchmark for the study of methods of human detection on depth map.The depth map is smoothed with pixel filtering and context filtering so as to reduce particulate noise.Then,the performance of five image features and a new feature is studied and compared for human detection on the dataset through theoretic analysis and simulation experiments.Results show that the new feature outperforms other descriptors.

A Multisource Contour Matching Method Considering the Similarity of Geometric Features

The existing multi-source contour matching studies have focused on the matching methods with consideration of topological relations and similarity measurement based on spatial Euclidean distance,while it is lack of taking the contour geometric features into account,which may lead to mismatching in map boundaries and areas with intensive contours or extreme terrain changes.In light of this,it is put forward that a matching strategy from coarse to precious based on the contour geometric features.The proposed matching strategy can be described as follows.Firstly,the point sequence is converted to feature sequence according to a feature descriptive function based on curvature and angle of normal vector.Then the level of similarity among multi-source contours is calculated by using the longest common subsequence solution.Accordingly,the identical contours could be matched based on the above calculated results.In the experiment for the proposed method,the reliability and efficiency of the matching method are verified using simulative datasets and real datasets respectively.It has been proved that the proposed contour matching strategy has a high matching precision and good applicability.

Topological distance-constrained feature descriptor learning model for vessel matching in coronary angiographies

Background Feature matching technology is vital to establish the association between virtual and real objects in virtual reality and augmented reality systems.Specifically,it provides them with the ability to match a dynamic scene.Many image matching methods,of which most are deep learning-based,have been proposed over the past few decades.However,vessel fracture,stenosis,artifacts,high background noise,and uneven vessel gray-scale make vessel matching in coronary angiography extremely difficult.Traditional matching methods perform poorly in this regard.Methods In this study,a topological distance-constrained feature descriptor learning model is proposed.This model regards the topology of the vasculature as the connection relationship of the centerline.The topological distance combines the geodesic distance between the input patches and constrains the descriptor network by maximizing the feature difference between connected and unconnected patches to obtain more useful potential feature relationships.Results Matching patches of different sequences of angiographic images are generated for the experiments.The matching accuracy and stability of the proposed method is superior to those of the existing models.Conclusions The proposed method solves the problem of matching coronary angiographies by generating a topological distance-constrained feature descriptor.

Target classification using SIFT sequence scale invariants

On the basis of scale invariant feature transform（SIFT） descriptors,a novel kind of local invariants based on SIFT sequence scale（SIFT-SS） is proposed and applied to target classification.First of all,the merits of using an SIFT algorithm for target classification are discussed.Secondly,the scales of SIFT descriptors are sorted by descending as SIFT-SS,which is sent to a support vector machine（SVM） with radial based function（RBF） kernel in order to train SVM classifier,which will be used for achieving target classification.Experimental results indicate that the SIFT-SS algorithm is efficient for target classification and can obtain a higher recognition rate than affine moment invariants（AMI） and multi-scale auto-convolution（MSA） in some complex situations,such as the situation with the existence of noises and occlusions.Moreover,the computational time of SIFT-SS is shorter than MSA and longer than AMI.

Summed volume region selection based three-dimensional automatic target recognition for airborne LIDAR

Airborne LIDAR can flexibly obtain point cloud data with three-dimensional structural information,which can improve its effectiveness of automatic target recognition in the complex environment.Compared with 2D information,3D information performs better in separating objects and background.However,an aircraft platform can have a negative influence on LIDAR obtained data because of various flight attitudes,flight heights and atmospheric disturbances.A structure of global feature based 3D automatic target recognition method for airborne LIDAR is proposed,which is composed of offline phase and online phase.The performance of four global feature descriptors is compared.Considering the summed volume region(SVR) discrepancy in real objects,SVR selection is added into the pre-processing operations to eliminate mismatching clusters compared with the interested target.Highly reliable simulated data are obtained under various sensor’s altitudes,detection distances and atmospheric disturbances.The final experiments results show that the added step increases the recognition rate by above 2.4% and decreases the execution time by about 33%.

Multi-Scale PIIFD for Registration of Multi-Source Remote Sensing Images

This paper aims at providing multi-source remote sensing images registered in geometric space for image fusion.Focusing on the characteristics and differences of multi-source remote sensing images,a feature-based registration algorithm is implemented.The key technologies include image scale-space for implementing multi-scale properties,Harris corner detection for keypoints extraction,and partial intensity invariant feature descriptor(PIIFD)for keypoints description.Eventually,a multi-scale Harris-PIIFD image registration algorithm framework is proposed.The experimental results of fifteen sets of representative real data show that the algorithm has excellent,stable performance in multi-source remote sensing image registration,and can achieve accurate spatial alignment,which has strong practical application value and certain generalization ability.

Robust multi-sensor image matching based on normalized self-similarity region descriptor

Multi-modal image matching is crucial in aerospace applications because it can fully exploit the complementary and valuable information contained in the amount and diversity of remote sensing images.However,it remains a challenging task due to significant non-linear radiometric,geometric differences,and noise across different sensors.To improve the performance of heterologous image matching,this paper proposes a normalized self-similarity region descriptor to extract consistent structural information.We first construct the pointwise self-similarity region descriptor based on the Euclidean distance between adjacent image blocks to reflect the structural properties of multi-modal images.Then,a linear normalization approach is used to form Modality Independent Region Descriptor(MIRD),which can effectively distinguish structural features such as points,lines,corners,and flat between multi-modal images.To further improve the matching accuracy,the included angle cosine similarity metric is adopted to exploit the directional vector information of multi-dimensional feature descriptors.The experimental results show that the proposed MIRD has better matching accuracy and robustness for various multi-modal image matching than the state-of-the-art methods.MIRD can effectively extract consistent geometric structure features and suppress the influence of SAR speckle noise using non-local neighboring image blocks operation,effectively applied to various multi-modal image matching.

Object tracking method based on joint global and local feature descriptor of 3D LIDAR point cloud

To fully describe the structure information of the point cloud when the LIDAR-object distance is long,a joint global and local feature(JGLF)descriptor is constructed.Compared with five typical descriptors,the object recognition rate of JGLF is higher when the LIDAR-object distances change.Under the situation that airborne LIDAR is getting close to the object,the particle filtering(PF)algorithm is used as the tracking frame.Particle weight is updated by comparing the difference between JGLFs to track the object.It is verified that the proposed algorithm performs 13.95%more accurately and stably than the basic PF algorithm.

A novel dense descriptor based on structure tensor voting for multi-modal image matching

Automatic and robust matching of multi-modal images can be challenging owing to the nonlinear intensity differences caused by radiometric variations among these images.To address this problem,a novel dense feature descriptor and improved similarity measure are proposed for enhancing the matching performance.The proposed descriptor is built on a voting scheme of structure tensor that can effectively capture the geometric structural properties of images.It is not only illumination and contrast invariant but also robust against the degradation caused by significant noise.Further,the similarity measure is improved to adapt to the reversal of orientation caused by the intensity inversion between multi-modal images.The proposed dense feature descriptor and improved similarity measure enable the development of a robust and practical templatematching algorithm for multi-modal images.We verify the proposed algorithm with a broad range of multi-modal images including optical,infrared,Synthetic Aperture Radar(SAR),digital surface model,and map data.The experimental results confirm its superiority to the state-of-the-art methods.

Applying rotation-invariant star descriptor to deep-sky image registration

Image registration is a critical process of many deep-sky image processing applications. Image registration methods include image stacking to reduce noise or achieve long exposure effects within a short exposure time, image stitching to extend the field of view, and atmospheric turbu- lence removal. The most widely used method for deep-sky image registration is the triangle- or polygon-based method, which is both memory and computation intensive. Deepsky image registration mainly focuses on translation and rotation caused by the vibration of imaging devices and the Earth＇s rotation, where rotation is the more difficult problem. For this problem, the best method is to find corresponding rotation-invariant features between different images. In this paper, we analyze the defects introduced by applying rotation-invariant feature descriptors to deep-sky image reg- istration and propose a novel descriptor. First, a dominant orientation is estimated from the geometrical relationships between a described star and two neighboring stable stars. An adaptive speeded-up robust features （SURF） descriptor is then constructed. During the construction of SURF, the local patch size adaptively changes based on the described star size. Finally, the proposed descriptor is formed by fusing star properties, geometrical relationships, and the adaptive SURF. Extensive experiments demonstrate that the proposed descriptor successfully addresses the gap resulting from applying the traditional feature-based method to deep-sky image registration and performs well compared to state-of-the-art descriptors.