Subspace clustering is widely used to find clusters in different subspaces within a dataset. Autoencoders are popular deep subspace clustering methods using feature extraction and dimensional reduction. However, neural networks are vulnerable to overfitting, and therefore have limited potential for unsupervised subspace clustering. This paper proposes a deep multi-view subspace clustering network with feature boosting module to successfully extract meaningful features in different views and to fuse multi-view representations in a complementary manner for enhanced clustering results. The multi-view boosting provides the robust features for unsupervised clustering by emphasizing the features and removing the redundant noise. Quantitative and qualitative analysis on various benchmark datasets verifies that the proposed method outperforms state-of-the-art subspace clustering methods.

Photo style transfer aims to change the style of a given photo to a reference style image with the constraint by retaining the broad and faithful conservation of the content of the input image. Most previous algorithms still have challenging issues on how to exactly extract and represent the style of the image to avoid the interruption of human visual perception. In this paper, we present a texture preserving photo style transfer algorithm by separating the input image into texture and structure and then applying the deep structure style transfer network to effectively change the extracted style characteristics of the structure. The texture preserving photo style transfer overcomes the main drawback of the previous approaches like distortion and saturation of the boundary of the objects. The quantitative and qualitative experimental results including user study prove that the proposed photo style transfer is universally applicable comparing to remarkable previous approaches.

Robust vanishing point estimation has been widely applied to various applications in the field of computer vision and pattern recognition for robotics, advanced driver assistance systems, and autonomous driving vehicles. The major challenge for vanishing point detection lies in line segments, spurious vanishing candidate removal, and clustering for refinement. Recent vanishing point detection approaches have attempted to reduce the computational complexity involved with voting processes using optimized voter selection strategies to identify the vanishing point from line segments. This paper proposes a novel vanishing point detection method to select robust candidates, applying optimized minimum spanning tree-based clustering of the vanishing point candidates by analyzing the lines within a unit sphere domain. The proposed scheme was applied to an open database that included illumination, partial occlusion, and viewpoint changes to validate robustness without prior scene information.

Text-based retrieval systems have been popular, but content-based retrieval systems have gained widespread acceptance in recent years to directly retrieve diverse media based on their visual content, such as color, texture, and shape. Among many content-based retrieval systems, sketch-based media retrieval systems have attracted attention recently with the proliferation of tablet PCs and smart mobile devices. Sketch-based retrieval requires the user to draw a freehand sketch query, but freehand drawing can be challenging for those with limited drawing skills. This degrades retrieval performance, since successful retrieval depends on the quality of the sketch query image drawn by the user. To address this issue, we propose a real-time stroke guidance for freehand sketch retrieval that continuously displays next-stroke shadow sketches on the canvas based on the user's step-by-step partial strokes. We train a stroke guidance network that learns the mapping between the step-wise stroke relations to predict the user's next stroke. The proposed stroke guidance for freehand sketch retrieval system runs on a five step next-stroke prediction model that identifies candidate next-stroke sketches from a database of millions of sketches. The system retrieves variable number of sketch object classes at different drawing stages. During the initial sketching stage, diverse drawing possibilities are covered by retrieving multiple sketch classes; as the sketching progresses, the intended sketch class is narrowed down to one. Deep binary hashing is employed for efficient similarity matching of relevant next-stroke sketches. We extend the Google QuickDraw dataset to create a five step sketch stroke database. Qualitative and quantitative experiments are conducted to verify the effectiveness of the proposed system, which can be utilized for drawing guidance, tracing, and sketch retrieval. Tracing refers to the act of copying the shadowed line of a guiding image by drawing over its lines.

Structure-preserving image decomposition separates a given image into structure and texture by smoothing the image, simultaneously preserving or enhancing image edges. The well-studied problem of image decomposition is applied to various areas, such as image smoothing, detail enhancement, non-photorealistic rendering, image artistic rendering, and high-dynamic-range compression. In this paper, we propose a fast algorithm for structure-preserving image decomposition that adopts total variation (TV) minimization to the moving least squares (MLS) method with non-local weights, called structure adaptive TV (SATV) minimization. MLS with non-local weights provides high accuracy approximation that is robust to noise, and allows a fast convergence with TV regularization term. As a result, our proposed SATV preserves the dominant structure while flattening fine-scale details. The experimental results show that the SATV minimization algorithm provides faster and more robust image decomposition than the well-known previous approaches. We demonstrate the usefulness of our algorithm by presenting successful applications in image smoothing and detail enhancement.

While the performance of sentiment classification has steadily risen through the introduction of various feature-based methods and distributed representation-based approaches, less attention was given to the qualitative aspect of classification, for instance, the identification of useful words in individual opinion texts. We present an approach using set operations for identifying useful words for sentiment classification, and employ truncated singular value decomposition (SVD), a classic low-rank matrix decomposition technique for document retrieval, in order to tackle the issue of both synonymy and noise removal. The sentiment classification performance of our approach, which concatenates three kinds of features, outperforms the existing word-based and distributed word representation-based methods and is comparable to the existing state of the art distributed document representation-based approaches.

A technical method regarding to the improvement of transmission capacity of an optical wireless orthogonal frequency division multiplexing (OFDM) link based on a visible light emitting diode (LED) is proposed in this paper. An original OFDM signal, which is encoded by various multilevel digital modulations such as quadrature phase shift keying (QPSK), and quadrature amplitude modulation (QAM), is converted into a sparse one and then compressed using an adaptive sampling with inverse discrete cosine transform, while its error-free reconstruction is implemented using a L1-minimization based on a Bayesian compressive sensing (CS). In case of QPSK symbols, the transmission capacity of the optical wireless OFDM link was increased from 31.12 Mb/s to 51.87 Mb/s at the compression ratio of 40 %, while It was improved from 62.5 Mb/s to 78.13 Mb/s at the compression ratio of 20 % under the 16-QAM symbols in the error free wireless transmission (forward error correction limit: bit error rate of 10-3).

The 3D object retrieval systems receive great concerns in the fields of pattern recognition and computer graphics because of their diverse applications. Traditional approaches of view-based 3D object retrieval have focused on finding descriptors which can efficiently represent the specific geometric information of the 3D object. By combining the local and the global features in order to improve the performance of 3D object retrieval, we propose a sparse coding based feature optimization technique using the hybrid gradient features of the projected images from 3D object. Experimental results show the effectiveness of our proposed approach.

3D Object Retrieval is an important field of research with many application possibilities. One of the main goals in this research is the development of discriminative methods for similarity search. The descriptor-based approach to date has seen a lot of research attention, with many different extraction algorithms proposed. In previous work, we have introduced a simple but effective scheme for 3D model retrieval based on a spatially fixed combination of 3D object fragment descriptors. In this work, we propose a novel flexible combination scheme based on finding the best matching fragment descriptors to use in the combination. By an exhaustive experimental evaluation on established benchmark data we show the capability of the new combination scheme to provide improved retrieval effectiveness. The method is proposed as a versatile and inexpensive method to enhance the effectiveness of a given global 3D descriptor approach.

In this paper, we present a markerless 3D motion capture system based on a volume reconstruction technique of non rigid bodies. It depicts a new approach for pose estimation in order to fit an articulated body model into the captured real-time information. We aim at analyzing athlete's movements in real-time within a 3D interactive graphics system. The paper addresses recent trends in vision based analysis and its fusion with 3D interactive computer graphics. Hence, the proposed system presents new methods for the 3D reconstruction of human body parts from calibrated multiple cameras based on voxel carving techniques and a 3D pose estimation methodology using Pseudo-Zernike Moments applied to an articulated human body model. Several algorithms have been designed for the deployment within a GPGPU environment allowing us to calculate several principle process steps from segmentation and reconstruction to volume optimization in real-time.

S.3740-3743

The number of available 3D models in various areas increase steadily. Effective methods to search for those 3D models by content, rather than textual annotations, are crucial. For this purpose, we propose a new approach for content based 3D model retrieval by hand-drawn sketch images. This approach to retrieve visually similar mesh models from a large database consists of three major steps: (1) suggestive contour renderings from different viewpoints to compare against the user drawn sketches; (2) descriptor computation by analyzing diffusion tensor fields of suggestive contour images or the query sketch respectively; (3) similarity measurement to retrieve the models and the most probable view-point from which a model was sketched. Our proposed sketch based 3D model retrieval system is very robust against variations of shape, pose or partial occlusion of the user draw sketches. Experimental results are presented and indicate the effectiveness of our approach for sketch-based 3D mode retrieval.

The understanding of object's kinematic structure is one of main challenges in the area of computer vision. Especially, skeleton of deformable objects, which is familiar with human visual perception, visualizes its characteristic using few data. This paper describes an efficient approach for automatic skeleton extraction and its splitting in the space of diffusion tensor fields, which are generated from normalized gradient vector flow fields of a given image. Our method is based on two steps: Skeleton extraction using second order diffusion tensor fields, Splitting skeleton using dissimilarity measure between neighbor elements. The evaluation proofs the efficiency of our technique which might be applied to object retrieval, pose estimation and action recognition, object registration and visualization.

This paper addresses a methodology on how users might interact with objects which have been reconstructed from uncalibrated multiple images. Tracking the user's eye movement for Human-Computer Interaction in an augmented reality environment provides a convenient, natural, and highbandwidth source for navigating and zooming in/out of the 3D reconstructed object. The calculated 3D eye position, which is detected by multiple cues from a stereo camera, is synchronized with the 3D position of 3D reconstructed object. Our proposed method of image based 3D modelling is based on voxel carving with photo consistency check. Experimental results show that our proposed interaction methodology using a new solution of 3D eye tracking can be successfully applied for useful tools in context aware applications and improve its usability.

Speaker segmentation is an important task in multi-party conversations. Overlapping speech poses a serious problem in segmenting audio into speaker turns. We propose an audio-visual speech separation system consisting of an array microphone with eight sensors and an omnidirectional color camera. Multiple concurrent speeches are segmented by fusing the two heterogeneous sensors. Each segmented speech is further enhanced by a linearly constrained minimum variance beamformer. Regardless of co-existing wide-band sound sources and pictures of human in a reverberant environment the proposed system effectively separates multiple target speeches.