Ken Chen

Asymmetric Coding of Multi-View Video Plus Depth Based 3-D Video for View Rendering

The recent years have witnessed three-dimensional (3-D) video technology to become increasingly popular, as it can provide high-quality and immersive experience to end users, where view rendering with depth-image-based rendering (DIBR) technique is employed to generate the virtual views. Distortions in depth map may induce geometry changes in the virtual views, and distortions in texture video may be propagated to the virtual views. Thus, effective compression of both texture videos and depth maps is important for 3-D video system. From the perspective of bit allocation, asymmetric coding of the texture videos and depth maps is an effective way to get the optimal solution of 3-D video compression and view rendering problems. In this paper, a novel asymmetric coding method of multi-view video plus depth (MVD) based 3-D video is proposed on purpose of providing high-quality view rendering. In the proposed method, two models are proposed to characterize view rendering distortion and binocular suppression in 3-D video. Then, an asymmetric coding method of MVD-based 3-D video is proposed by combining two models in encoding framework. Finally, a chrominance reconstruction algorithm is presented to achieve accurate reconstruction. Experimental results show that compared with other methods, the proposed method can obtain higher performance of view rendering under the total bitrate constraint. Moreover, the perceptual visual quality of 3-D video is almost unaffected with the proposed method.

Stereoscopic visual attention-based regional bit allocation optimization for multiview video coding

We propose a Stereoscopic Visual Attention- (SVA-) based regional bit allocation optimization for Multiview Video Coding (MVC) by the exploiting visual redundancies from human perceptions. We propose a novel SVA model, where multiple perceptual stimuli including depth, motion, intensity, color, and orientation contrast are utilized, to simulate the visual attention mechanisms of human visual system with stereoscopic perception. Then, a semantic region-of-interest (ROI) is extracted based on the saliency maps of SVA. Both objective and subjective evaluations of extracted ROIs indicated that the proposed SVA model based on ROI extraction scheme outperforms the schemes only using spatial or/and temporal visual attention clues. Finally, by using the extracted SVA-based ROIs, a regional bit allocation optimization scheme is presented to allocate more bits on SVA-based ROIs for high image quality and fewer bits on background regions for efficient compression purpose. Experimental results on MVC show that the proposed regional bit allocation algorithm can achieve over % bit-rate saving while maintaining the subjective image quality. Meanwhile, the image quality of ROIs is improved by u2009dB at the cost of insensitive image quality degradation of the background image.

Image profile area calculation based on circular sample measurement calibration

A practical approach of measurement calibration is presented for obtaining the true area of the photographed objects projected in the 2-D image scene. The calibration is performed using three circular samples with given diameters. The process is first to obtain the ratio mm/pixel in two orthogonal directions, and then use the obtained ratios with the total number of pixels scanned within projected area of the object of interest to compute the desired area. Compared the optically measured areas with their corresponding true areas, the results show that the proposed method is quite encouraging and the relevant application also proves the approach adequately accurate.

A novel rate control technique for asymmetric-quality stereoscopic video

Three-dimensional (3D) video technology is becoming increasingly popular, as it can provide stereoscopic perception and immersive experience to end users. Some asymmetric stereoscopic video coding methods had been developed by utilizing binocular psycho-visual redundancy. However, rate control problem had not been well considered in these methods to further control the quality of stereoscopic video. In this paper, a novel rate control method is proposed for asymmetric-quality stereoscopic video. In order to model the asymmetric-quality stereoscopic video, we use a fixed threshold to quantize the binocular psycho-visual redundancy and establish the relationship between the distortion and quantization for the asymmetric-quality stereoscopic video. Then, a rate control method is engineered in stereoscopic video coding to control the rate and quality of left and right views. Experimental results show that the proposed method can accurately control the rate and quality of stereoscopic video, while having a lower computational complexity1.

A Bhattacharyya-factor based Camshift application for video fast mobile target tracking

A Bhattacharyya-factor based scaling-down search window is proposed for tracking fast moving video target. Due to tracking errors arising from the limitation of initial search window size, the tracking result is prone to being trapped with local optimization when Camshift is directly applied. Aiming at the drawback identified, the author presents an approach in which the initial search window mapping is determined using Bhattacharyya factor in the scaled-down window area with color characteristics mapping of video target of interest, followed by utilizing Camshift algorithm to finally detecting the precise target position. Judging by the tracking precision defined, the tests are conducted on the real-world video sequence, suggesting the satisfactory tracking quality and its feasibility.

Effect of multi-hidden-layer structure on performance of BP neural network: Probe

As a multi-layer forwarding network, the back propagation neural network (BPNN) with manifold derived structures has been most widely used in artificial intelligence applications. Based on the given non-linear system and the BPNNs of varying internal structures, this paper quantitatively reports the findings in the correlation between the number of hidden layers and the BPNN performance. The selection of learning rate is also investigated using the 3-layer BPNN and the same non-linear system. Through the simulation results in this probe it finds that the BPNN performance is not improved notably or even degraded with the increase of hidden layers, and 3-layer (or 1-1-1) BPNN is identified as the best performer.

Online parameter based Kalman filter precision evaluation method for video target tracking

In this paper, we propose a new method for evaluating the precision of Kalman filter in video target tracking. Firstly, the performance evaluation model is presented which is different from the usual method that we handle the video. Then the parameter that used to assess the performance of the tracking algorithms is defined and utilized in effective evaluation procedure. In the end, experiments are conducted to validate the proposed approach, which indicates that the proposed performance evaluation norm can not only reflect at which part the algorithm performs well or not, but also can perceive the precision of Kalman filter in target tracking.

A Meanshift-based imbedded computer vision system design for real-time target tracking

Visual tracking is among the newly burgeoning subjects along with advancement of computer vision systems, and embedded tracking system is a conjunction of hardware-embedded communications and control system, as well as software-embedded tracking system. In this paper, the Meanshift algorithm implementing visual tracking is introduced. The integral hardware design is based on the TE6410 development board as the platform, and on WinCE as the operating system. The software design involves coding for the embedded operating system and applying the multi-thread technology. Combined with a drive-module of servomotor, a video camera and a VGA display, the tracking test is conducted, showing that within the limited number of iterations, the Meanshift is capable of visual tracking in a real-time manner using the developed imbedded system.

KF vs. PF Performance Quality Observed from Stochastic Noises Statistics and Online Covariance Self-adaptation

The workability of Kalman filter is explored in the perspective of various stochastic noise statistics. The Gaussage is defined and applied at different levels in both Kalman filtering and particle filtering to evaluate the performance quality. Two parameters, DOM and DOD, are introduced and used for checking the consistence between the assumed stochastic noise covariance in Kalman filter and the truly existing covariance. Based on the proposed parameters, the method of online covariance adaptation is engineered, which is aimed to ultimately achieve the optimal Kalman performance quality.

Video motion tracking using enhanced particle filtering with Mean-shift

Some weaknesses of particle filtering has been have been identified from an application perspective. This paper proposes a trial approach to tackle the problems of exceeding number of particles required for sampling, low particle efficiency, and compromised particle diversity resulting from resampling. The method combines particle filtering with Mean-shift, which is used to further optimize the sampled particles, thus significantly reducing the number of particles while retaining the particle diversity. The Bhattacharyya factor is induced to determine the importance weighting of particle. The test results exhibit that the proposed method can perform the robust tracking in the face of high mobility, partial occlusion, and limited rotation.