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 o...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.展开更多
To solve the problem of perception failure due to vehicle occlusion in urban areas,this paper will review and analyze many dynamic detection and tracking methods proposed in recent years in detail.The main idea of the...To solve the problem of perception failure due to vehicle occlusion in urban areas,this paper will review and analyze many dynamic detection and tracking methods proposed in recent years in detail.The main idea of the first representative study is to use laser beams and geometric figures for rigid matching.They are very effective in open areas,but because there is no information on the surface,the algorithm cannot work when there is occlusion.Earlier,researchers put forward a soft-assignment method by building a“four-rectangle measurement model”,and it was found that probabilistic modeling is better than binary matching;now,a laser ray does not need to hit an object,and the matching probability of local sparse point clouds can be calculated.A good way to improve the efficiency of tracking is the scaled sequence algorithm.The first problem of this algorithm is that it involves a high-dimensional state search.The detection rate of severe occlusion in the KITTI dataset for the new model is higher than that of traditional methods.As shown in the above case,although this method is relatively stable in the face of missing values,it has some defects;that is,it is too sensitive to the initial prior and may be trapped in a local optimum under difficult conditions.Due to the shortcomings of the previous models,such as fixed dimensions and an inability to address all kinds of traffic flexibly,it is hoped that future research will be based on multi-scale adaptive geometric models to achieve all-weather sensing and all-conditions operation.Vehicle occlusion,Likelihood Field Model,Scaling sequence algorithm,self-motion compensation.展开更多
There are many challenges to achieving quality multiscalar displays of vector data in geographical information system(GIS).Acknowledging and making use of the differences between traditional cartographic generalizatio...There are many challenges to achieving quality multiscalar displays of vector data in geographical information system(GIS).Acknowledging and making use of the differences between traditional cartographic generalization and multiscalar display in GIS,this paper focuses on novel approaches to the definition of small objects and the establishment of scale sequence.The ease of implementation and efficacy of the solutions proposed are exemplified and analyzed.Possibilities for further research into the multiscalar display of vector data in GIS are thereby suggested.展开更多
Scale is the basic attribute for expressing anddescribing spatial entity and phenomena. It offerstheoretical significance in the study of gully structureinformation, variable characteristics of watershed mor-phology, ...Scale is the basic attribute for expressing anddescribing spatial entity and phenomena. It offerstheoretical significance in the study of gully structureinformation, variable characteristics of watershed mor-phology, and development evolution at different scales.This research selected five different areas in China's LoessPlateau as the experimental region and used DEM data atdifferent scales as the experimental data. First, the changerule of the characteristic parameters of the data at differentscales was analyzed. The watershed structure informationdid not change along with a change in the data scale. Thiscondition was proven by selecting indices of gullybifurcation ratio and fractal dimension as characteristicparameters of watershed structure information. Then, thechange rule of the characteristic parameters of gullystructure with different analysis scales was analyzed bysetting the scale sequence of analysis at the extractiongully. The gully structure of the watershed changed withvariations in the analysis scale, and the change rule wasobvious when the gully level changed. Finally, the changerule of the characteristic parameters of the gully structure atdifferent areas was analyzed. The gully fractal dimensionshowed a significant numerical difference in differentareas, whereas the variation of the gully branch ratio wassmall. The change rule indicated that the developmentdegree of the gully obviously varied in different regions,but the morphological structure was basically similar.展开更多
The periodic multiresolution analysis (PMRA) and the periodic frame multiresolution analysis (PFMRA) provide a general recipe for the construction of periodic wavelets and periodic wavelet frames, respectively. Th...The periodic multiresolution analysis (PMRA) and the periodic frame multiresolution analysis (PFMRA) provide a general recipe for the construction of periodic wavelets and periodic wavelet frames, respectively. This paper addresses PFMRAs by the introduction of the notion of spectrum sequence. In terms of spectrum sequences, the scaling function sequences generating a normalized PFMRA are characterized; a characterization of the spectrum sequences of PFMRAs is obtained, which provides a method to construct PFMRAs since its proof is constructive; a necessary and sufficient condition for a PFMRA to admit a single wavelet frame sequence is obtained; a necessary and sufficient condition for a PFMRA to be contained in a given PMRA is also obtained. What is more, it is proved that an arbitrary PFMRA must be contained in some PMRA. In the meanwhile, some examples are provided to illustrate the general theory.展开更多
In this paper, we study the properties of periodic multiresolution analysis, and present a complete characterization of the scaling function sequence, which enables us to construct a new scaling function sequence from...In this paper, we study the properties of periodic multiresolution analysis, and present a complete characterization of the scaling function sequence, which enables us to construct a new scaling function sequence from a given one. An application of the main results is given at the end.展开更多
基金supported by the National High Technology Research and Development Program (863 Program) (2010AA7080302)
摘要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.
摘要To solve the problem of perception failure due to vehicle occlusion in urban areas,this paper will review and analyze many dynamic detection and tracking methods proposed in recent years in detail.The main idea of the first representative study is to use laser beams and geometric figures for rigid matching.They are very effective in open areas,but because there is no information on the surface,the algorithm cannot work when there is occlusion.Earlier,researchers put forward a soft-assignment method by building a“four-rectangle measurement model”,and it was found that probabilistic modeling is better than binary matching;now,a laser ray does not need to hit an object,and the matching probability of local sparse point clouds can be calculated.A good way to improve the efficiency of tracking is the scaled sequence algorithm.The first problem of this algorithm is that it involves a high-dimensional state search.The detection rate of severe occlusion in the KITTI dataset for the new model is higher than that of traditional methods.As shown in the above case,although this method is relatively stable in the face of missing values,it has some defects;that is,it is too sensitive to the initial prior and may be trapped in a local optimum under difficult conditions.Due to the shortcomings of the previous models,such as fixed dimensions and an inability to address all kinds of traffic flexibly,it is hoped that future research will be based on multi-scale adaptive geometric models to achieve all-weather sensing and all-conditions operation.Vehicle occlusion,Likelihood Field Model,Scaling sequence algorithm,self-motion compensation.
基金Supported by the Knowledge Innovation Program of the Chinese Academy of Sciences(No.kzcx2-yw-345).
摘要There are many challenges to achieving quality multiscalar displays of vector data in geographical information system(GIS).Acknowledging and making use of the differences between traditional cartographic generalization and multiscalar display in GIS,this paper focuses on novel approaches to the definition of small objects and the establishment of scale sequence.The ease of implementation and efficacy of the solutions proposed are exemplified and analyzed.Possibilities for further research into the multiscalar display of vector data in GIS are thereby suggested.
摘要Scale is the basic attribute for expressing anddescribing spatial entity and phenomena. It offerstheoretical significance in the study of gully structureinformation, variable characteristics of watershed mor-phology, and development evolution at different scales.This research selected five different areas in China's LoessPlateau as the experimental region and used DEM data atdifferent scales as the experimental data. First, the changerule of the characteristic parameters of the data at differentscales was analyzed. The watershed structure informationdid not change along with a change in the data scale. Thiscondition was proven by selecting indices of gullybifurcation ratio and fractal dimension as characteristicparameters of watershed structure information. Then, thechange rule of the characteristic parameters of gullystructure with different analysis scales was analyzed bysetting the scale sequence of analysis at the extractiongully. The gully structure of the watershed changed withvariations in the analysis scale, and the change rule wasobvious when the gully level changed. Finally, the changerule of the characteristic parameters of the gully structure atdifferent areas was analyzed. The gully fractal dimensionshowed a significant numerical difference in differentareas, whereas the variation of the gully branch ratio wassmall. The change rule indicated that the developmentdegree of the gully obviously varied in different regions,but the morphological structure was basically similar.
基金Supported by the National Natural Science Foundation of China (Grant No. 10671008)supported by the Excellent Talents Foundation of Beijing, China (20051D0501022)PHR(IHLB)+1 种基金the Project-sponsored by SRF for ROCS, SEM of Chinathe Scientific Research Foundation for the Excellent Returned Overseas Chinese Scholars, Beijing
摘要The periodic multiresolution analysis (PMRA) and the periodic frame multiresolution analysis (PFMRA) provide a general recipe for the construction of periodic wavelets and periodic wavelet frames, respectively. This paper addresses PFMRAs by the introduction of the notion of spectrum sequence. In terms of spectrum sequences, the scaling function sequences generating a normalized PFMRA are characterized; a characterization of the spectrum sequences of PFMRAs is obtained, which provides a method to construct PFMRAs since its proof is constructive; a necessary and sufficient condition for a PFMRA to admit a single wavelet frame sequence is obtained; a necessary and sufficient condition for a PFMRA to be contained in a given PMRA is also obtained. What is more, it is proved that an arbitrary PFMRA must be contained in some PMRA. In the meanwhile, some examples are provided to illustrate the general theory.
摘要In this paper, we study the properties of periodic multiresolution analysis, and present a complete characterization of the scaling function sequence, which enables us to construct a new scaling function sequence from a given one. An application of the main results is given at the end.