Near-duplicate document image matching: A graphical perspective

A near-duplicate document image matching approach characterized by a graphical perspective is proposed in this paper. Document images are represented by graphs whose nodes correspond to the objects in the images. Consequently, the image matching problem is then converted to graph matching. To deal with the instability of object segmentation, a multi-granularity object tree is constructed for a document image. Each level in the tree corresponds to one possible object segmentation, while different levels are characterized by various object granularities. Some graphs can be generated from the tree and the objects associated with each graph may be of different granularities. Two graphs with the maximum similarity are found from the multi-granularity object trees of the two near-duplicate document images which are to be matched. The encouraging experimental results have demonstrated the effectiveness of the proposed approach.     

Efficient multilevel image segmentation through fuzzy entropy maximization and graph cut optimization

The fuzzy c-partition entropy has been widely adopted as a global optimization technique for finding the optimized thresholds for multilevel image segmentation. However, it involves expensive computation as the number of thresholds increases and often yields noisy segmentation results since spatial coherence is not enforced. In this paper, an iterative calculation scheme is presented for reducing redundant computations in entropy evaluation. The efficiency of threshold selection is further improved through utilizing the artificial bee colony algorithm as the optimization technique. Finally, instead of performing thresholding for each pixel independently, the presented algorithm oversegments the input image into small regions and uses the probabilities of fuzzy events to define the costs of different label assignments for each region. The final segmentation results is computed using graph cut, which produces smooth segmentation results. The experimental results demonstrate the presented iterative calculation scheme can greatly reduce the running time and keep it stable as the number of required thresholds increases. Quantitative evaluations over 20 classic images also show that the presented algorithm outperforms existing multilevel segmentation approaches.     

Graph regularized multiset canonical correlations with applications to joint feature extraction

Multiset canonical correlation analysis (MCCA) is a powerful technique for analyzing linear correlations among multiple representation data. However, it usually fails to discover the intrinsic geometrical and discriminating structure of multiple data spaces in real-world applications. In this paper, we thus propose a novel algorithm, called graph regularized multiset canonical correlations (GrMCCs), which explicitly considers both discriminative and intrinsic geometrical structure in multiple representation data. GrMCC not only maximizes between-set cumulative correlations, but also minimizes local intraclass scatter and simultaneously maximizes local interclass separability by using the nearest neighbor graphs on within-set data. Thus, it can leverage the power of both MCCA and discriminative graph Laplacian regularization. Extensive experimental results on the AR, CMU PIE, Yale-B, AT&T, and ETH-80 datasets show that GrMCC has more discriminating power and can provide encouraging recognition results in contrast with the state-of-the-art algorithms.     

HEGM: A hierarchical elastic graph matching for hand gesture recognition

A hierarchical scheme for elastic graph matching applied to hand gesture recognition is proposed. The proposed algorithm exploits the relative discriminatory capabilities of visual features scattered on the images, assigning the corresponding weights to each feature. A boosting algorithm is used to determine the structure of the hierarchy of a given graph. The graph is expressed by annotating the nodes of interest over the target object to form a bunch graph. Three annotation techniques, manual, semi-automatic, and automatic annotation are used to determine the position of the nodes. The scheme and the annotation approaches are applied to explore the hand gesture recognition performance. A number of filter banks are applied to hand gestures images to investigate the effect of using different feature representation approaches. Experimental results show that the hierarchical elastic graph matching (HEGM) approach classified the hand posture with a gesture recognition accuracy of 99.85% when visual features were extracted by utilizing the Histogram of Oriented Gradient (HOG) representation. The results also provide the performance measures from the aspect of recognition accuracy to matching benefits, node positions correlation and consistency on three annotation approaches, showing that the semi-automatic annotation method is more efficient and accurate than the other two methods.     

Depth-based complexity traces of graphs

In this paper we aim to characterize graphs in terms of a structural measure of complexity. Our idea is to decompose a graph into layered substructures of increasing size, and then to measure the information content of these substructures. To locate dominant substructures within a graph, we commence by identifying a centroid vertex which has the minimum shortest path length variance to the remaining vertices. For each graph a family of centroid expansion subgraphs is derived from the centroid vertex in order to capture dominant structural characteristics of the graph. Since the centroid vertex is identified through a global analysis of the shortest path length distribution, the expansion subgraphs provide a fine representation of a graph structure. We then show how to characterize graphs using depth-based complexity traces. Here we explore two different strategies. The first strategy is to measure how the entropies on the centroid expansion subgraphs vary with the increasing size of the subgraphs. The second strategy is to measure how the entropy differences vary with the increasing size of the subgraphs. We perform graph classification in the principal component space of the complexity trace vectors. Experiments on graph datasets abstracted from some bioinformatics and computer vision databases demonstrate the effectiveness and efficiency of the proposed graph complexity traces. Our methods are competitive to state of the art methods.     

Tool path generation framework for accurate manufacture of complex 3D sheet metal parts using single point incremental forming

The prototyping of complex sheet metal parts using single point incremental forming (SPIF) requires the generation of optimal tool paths and/or tool path sequences that ensure that the formed part is within geometric design specifications. The presence of a multitude of features on complex parts leads to multiple inaccuracy inducing phenomena occurring simultaneously due to interactions between the features. This paper proposes a network analysis methodology using topological conceptual graphs to capture the effects of different phenomena on the final accuracy of a sheet metal part manufactured by SPIF. Using this framework optimized tool paths can be generated that compensate for the inaccuracy inducing behavior. Tool path generation algorithms to create partial tool paths that account for the accuracy of specific features in the part based on the proposed framework are also presented. Finally, the creation of integrated tool paths maintaining complementarity between tool paths and desired continuity behavior using non-uniform cubic B-splines is illustrated. A number of case studies demonstrating the applicability of the integrated framework are discussed, where the maximum deviations in the part are significantly reduced and the average absolute deviations for the complete part are brought down to less than 0.5 mm.     

基于特征点跟踪的视频序列平面提取算法

平面结构是一种最为常见和重要的几何特征,平面提取在很多领域有着重要的应用。提出了一种毋须恢复摄像机三维信息、直接从视频序列中提取出平面的算法。先通过在视频相邻帧之间建立极线约束下的射影变换模型,来计算出由平面诱导的单应,而每一个单应都对应着场景中的一个平面;然后根据计算出的单应对视频的每一帧进行分割,进一步提取出视频中的平面结构。一系列真实视频序列的实验验证了本算法的有效性和稳定性。     

基于混合图像分割与梯度算法的发光物体图像重建技术研究

传统的图像重建算法存在光源分布不均以及噪声干扰等问题,导致图像重建效果差。针对该问题,提出了一种改进的混合图割算法和梯度算法的发光体图像重建技术。算法首先采用图像分割算法得到在未知先验条件的情况下的发光源情况;然后利用不同的梯度算法,根据重建状态得到发光源准确的分布情况;最后利用内部光源的多级网络提高计算速度和重建的准确性。仿真实验结果表明,本方法即使在存在检测噪声和模型结构误差的情况下,仍然能够得到很好的重建性能,具有较高的实际应用价值。     

基于分辨函数的极大团搜索算法

寻找极大团是几何图论极为重要的基础研究问题之一。将分辨函数模型与极大团性质结合,定义了顶点的极大团分辨函数、顶点关于某顶点子集的布尔映射函数,得到了一些与极大团相关的重要性质与定理,证明了图的极大团搜索问题可快捷自然地转换为相对简单的分辨函数表达式约束,为设计极大团搜索算法提供了一种有效的理论依据与求解途径。进而引入约简树构造方法设计了基于分辨函数的极大团搜索算法,最后通过给定无向连通图实例说明了算法的可行性与有效性。