Active Shape Models Using Scale Invariant Feature Transform

A new active shape models (ASMs) was presented, which is driven by scale invariant feature transform (SIFT) local descriptor instead of normalizing first order derivative profiles in the original formulation, to segment lung fields from chest radiographs. The modified SIFT local descriptor, more distinctive than the general intensity and gradient features, is used to characterize the image features in the vicinity of each pixel at each resolution level during the segmentation optimization procedure. Experimental results show that the proposed method is more robust and accurate than the original ASMs in terms of an average overlap percentage and average contour distance in segmenting the lung fields from an available public database.

A Note on Unification of Translational Shape Invariant Potential and Scaling Shape Invariant Potential

This article puts forward a general shape invariant potential, which includes the translational shape invariant potential and scaling shape invariant potential as two particular cases, and derives the set of linear differential equations for obtaining general solutions of the generalized shape invariance condition.

Exact Solution of Klein-Gordon Equation for Charged Particle in Magnetic Field with Shape Invariant Method

Based on the shape invariance property we obtain exact solutions of the three-dimensional relativistic Klein Gordon equation for a charged particle moving in the presence of a certain varying magnetic field, and we also show its non-relativistic limit.

Generation and classification of the translational shape-invariant potentials based on the analytical transfer matrix method

For the conventional translational shape-invariant potentials （TSIPs）, it has demonstrated that the phase contribution devoted by the scattered subwaves in the analytical transfer matrix quantization condition is integrable and independent of n. Based on this fact we propose a novel strategy to generate the whole set of conventional TSIPs and classify them into three types. The generating functions are given explicitly and the Morse potential is taken as an example to illustrate this strategy.

Identity for the Exponential-Type Molecule Potentials and the Supersymmetry Shape Invariance

The identity and the supersymmetry shape invariance for a class of exponential-type molecule potentials are studied by introducing a deformed five-parameter exponential-type potential (DFPEP) and via the multi-parameter deformations. It has been shown that the DFPEP is a shape-invariant potential with a translation of parameters. By making use of the shape invariance approach, the exact energy levels are determined for the bound states with zero angular momentum. A class of molecule potentials and their exact energy spectra for the zero angular momentum states are reduced from the DFPEP and a general energy spectrum formula, respectively. The interrelations for some molecule potentials are also discussed.

基于Shape Context和尺度不变特征变换的多模图像自动配准方法

提出了基于修正的尺度不变特征变换(SIFT)特征提取和Shape Context特征描述算子相结合的多模图像自动配准算法,该算法利用修正的SIFT算法提取多模图像中的特征点,然后采用Shape Context算子描述特征点,利用特征点周围区域边缘点的梯度方向形成特征向量。采用欧氏距离作为匹配标准对多模图像中特征点进行初始匹配,然后通过RANSAC算法消除误匹配的特征点对,并采用最小二乘法计算仿射变换参数,最后通过仿射变换和双线性插值实现图像配准。对红外图像和可见光图像的配准实验结果表明了本算法的有效性和稳定性。

Shape matching and retrieval based on multiple feature descriptors

A lot of 3D shape descriptors for 3D shape retrieval have been presented so far. This paper proposes a new mechanism, which employs several existing global and local 3D shape descriptors as input. With the sparse theory, some descriptors which play the most important role in measuring similarity between query model and the model in the dataset are selected automatically and an affinity matrix is constructed. Spectral clustering method can be implemented to this affinity matrix. Spectral embedding of this affinity matrix can be applied to retrieval, which integrating almost all the advantages of selected descriptors. In order to verify the performance of our approach, we perform experimental comparisons on Princeton Shape Benchmark database. Test results show that our method is a pose-oblivious, efficient and robustness method for either complete or incomplete models.

Shape classification based on singular value decomposition transform

In this paper, a new shape classification system based on singular value decomposition (SVD) transform using nearest neighbour classifier was proposed. The gray scale image of the shape object was converted into a black and white image. The squared Euclidean distance transform on binary image was applied to extract the boundary image of the shape. SVD transform features were extracted from the the boundary of the object shapes. In this paper, the proposed classification system based on SVD transform feature extraction method was compared with classifier based on moment invariants using nearest neighbour classifier. The experimental results showed the advantage of our proposed classification system.

Bound states of the Schrdinger equation for the Pschl-Teller double-ring-shaped Coulomb potential

Poschl-Teller double-ring-shaped Coulomb （PTDRSC） potential, the Coulomb potential surrounded by PSschl- Teller and double-ring-shaped inversed square potential, is put forward. In spherical polar coordinates, PTDRSC potential has supersymmetry and shape invariance in φ,θ and τ coordinates. By using the method of supersymmetry and shape invariance, exact bound state solutions of Schr6dinger equation with PTDRSC potential are presented. The normalized φ,θ angular wave function expressed in terms of Jacobi polynomials and the normalized radial wave function expressed in terms of Laguerre polynomials are presented. Energy spectrum equations are obtained. Wave function and energy spectrum equations of the system are related to three quantum numbers and parameters of PTDRSC potential. The solutions of wave functions and corresponding eigenvalues are only suitable for the PTDRSC potential.

Evolution of crescent-shaped sand dune under the influence of injected sand flux:scaling law and wind tunnel experiment

This paper studies the evolution of crescent-shaped dune under the influence of injected flux. A scaling law and a wind tunnel experiment are carried out for comparison. The experiment incorporates a novel image processing algorithm to recover the evolutionary process. The theoretical and experimental results agree well in the middle stage of dune evolution, but deviate from each other in the initial and final stages, suggesting that the crescent-shaped dune evolution is intrinsically scale-variant and that the crescent shape breaks down under unsaturated condition.

基于形状不变矩的图像检索算法的研究

描述了一种基于Hu形状不变矩的图像全局形状特征提取方法和算法IMS。实验结果表明 ,使用IMS。算法提取的形状特征向量具有对平移、旋转和尺度变化的不变性 。

一种新的几何特征形状描述子

通过对平移旋转和伸缩变换下的不变量进行研究,证明了形状的长轴端点和重心是该变换下的不变量,提出了长轴两端点与重心所成的夹角这一几何特征的形状描述子,该形状描述子具有平移旋转和伸缩变换不变性,理论分析和实验结果表明该描述子简单高效,具有较好的形状匹配结果。

基于改进形状上下文特征的二值图像检索

提出了改进的形状上下文算法以克服传统的形状上下文算法不具备旋转不变性这一缺点。该算法利用找寻包含采样点数最多的角度区间的方式改变图像角度,对相对应的区域进行比较,并计算匹配代价,从而为形状上下文加入旋转不变性。为提高运算速度,算法也引入了剪枝方法,解决了进行直方图距离计算时遍历采样点的问题。实验显示,本文的算法在公开数据库上测试得到的精确度召回率(PR)曲线与郑提出算法的PR曲线性能接近,但是计算速度较其提升了近1倍;与传统的形状上下文算法相比,提出算法的PR曲线更为优越,且检索精度有较大提高。因此,提出的算法综合检索性能更好,能够有效地的应用于二值图像检索领域。

基于内容的图像检索技术研究

分析了基于内容的图像检索的关键技术以及各个阶段的研究特点,重点分析了基于颜色和基于形状的图像检索技术。指出了目前研究中存在的主要问题并提出了今后的研究方向。

GFO和不变矩实现基于形状的图像检索新方法

提出了一种新的基于形状的图像检索方法 .首先利用广义模糊算子对图像进行边缘检测 ,以检测出的边缘映射图计算不变矩来代表边缘图像的形状信息 ,从而完成对图像的自动索引 ;其次 ,重新定义了相似度测度 ,提高了匹配速度 ;最后 ,设定纵横比辅助搜索 ,以提高检索精度 .实验结果表明 ,此方法在一定程度上排除了与形状无关信息的影响 ,不但简单有效 ,而且搜索速度快 ,具有一定的实用意义 .

基于多特征的杂草逆向定位方法与试验

提出了一种基于多特征的杂草逆向定位方法。以田间作物作为研究对象,将多目标杂草定位问题转换为单目标的作物定位问题。采用作物叶片HU不变矩与形状特征的识别准确定位出每一株作物,然后基于颜色特征将作物区域以外的绿色植物均认定为杂草。设计了一款小型杂草定位装置,并应用在宽幅喷药机上。田间试验结果表明,在喷药机工作速度为5 km/h时,该系统对于大豆田间杂草识别的准确率为90%以上,较好地解决了杂草定位与精细喷洒农药问题。

用于模式识别的极半径不变矩

提出了用于目标物识别和分类的极半径不变矩 .对目标物进行分割后 ,先求出目标物的形心 ,进而求出极半径矩、归一化矩和归一化的中心矩 .在此基础上 ,给出了 5个具有平移、旋转和尺度变换不变性的特征量用于物体形状的识别 .文中给出了这些不变矩的特性 ,并给出了极半径不变矩和边界序列矩以及Hu提出的不变矩的实验比较结果 .该文提出的极半径不变矩 ,既可用于区域目标的识别 ,也可用于边界形状的识别 .

基于仿射不变矩的轴心轨迹自动识别方法

讨论了仿射不变矩的原理,通过改进仿射不变矩的积分因子,保证了积分因子在仿射变换下的不变性,将仿射不变矩应用于水电机组轴心轨迹特征的提取.通过计算轴心轨迹的仿射不变矩作为特征向量,结合人工神经网络方法,提出了一种高效的轴心轨迹自动识别新方法.对3类典型的轴心轨迹的自动识别进行了仿真试验,并与Hu氏不变矩方法的识别效果进行对比.试验结果表明:采用仿射不变矩方法的轴心轨迹识别率达到95%,明显高于Hu氏不变矩的识别率,仿射不变矩具有比Hu氏不变矩更好的轴心轨迹识别能力.

HARTMANN势的能谱和波函数

在量子化学中采用Ｈａｒｔｍａｎｎ势描绘环状分子，例如苯分子等。运用超对称性量子力学和形不变性的方法求解Ｈａｒｔｍａｎｎ势的能量本征值和本征函数。

基于Grassmann流形的仿射不变形状识别

传统的Kendall形状空间理论仅适用于相似变换,然而成像过程中目标发生的几何变形在更多情形时应该用仿射变换来刻画.基于Grassmann流形理论,本文分析了仿射不变形状空间的非线性几何结构,提出了基于Grassmann流形的仿射不变形状识别算法.算法首先对训练集中的每类形状分别计算形状均值和方差,进而在形状均值附近的切空间构建多变量正态分布;最后,根据测试形状的观测和先验形状模型求解测试形状的最大似然类,对形状进行贝叶斯分类.MPEG7形状数据库的实验结果表明,与传统Kendall形状分析中的基于Procrustean度量识别算法相比,本文识别算法具有明显优势;真实场景中的目标识别结果进一步表明,本文算法对仿射变形有更好的适应能力,在复杂场景下能以较高的后验概率辨识出目标类别.