Search

This paper utilizes the flexibility of microprogrammed controller with reloadable microcodes for developing reconfigurable parser for multiple purposes. Based on control-dataflow, with microprogrammed controller taken into consideration due to the nature of feedback control in parser, this paper proposes a reconfigurable parser through extracting control commonalities to form shared microinstructions so that the current architecture alleviates the cost in switching control signals for distinct purposes. We employ reconfigurable video coding as a case study to justify the advantages in reconfigurable parser with microprogrammed controller in comparison with finite state machine based controller. Using TSMC 0.18 μm CMOS technology at 108 MHz operating frequency, we reduce 8.93 % gate counts and increase throughput rate twice in comparison with individually implemented finite state machine based controller. We have demonstrated that microprogrammed controller is the trend of flexible architecture design for multiple purposes. Owing to high proportion of shared microinstruction, the higher saving ratio could be envision when multiple purposes are involved in the proposed reconfigurable parser, e.g., more video coding standards.

Melasma is the common hyperpigmentary skin disorder that can cause bright and dark brown macules and patches on sun-exposed areas of the skin owing to hyperactivity of epidermal melanocytes. Correspondingly, melanocyte detection and subsequent observation are principle procedures conducted by physicians. In this paper, we utilize methodology of contextual comparison based on intrinsic cell characteristics to detect epidermal melanocytes, which can provide consistent and accurate results for assisting the assessment of diagnosis. In addition, we propose the method of Computer-Aided Diagnosis (CAD) via moving window statistical analysis, which evaluate the relationship between detected melanocytes and the characteristics of the intercellular distribution in the Third Harmonic Generation (THG) image. The experimental results show the inconsistent distribution of cells around melanocytes in different disease categories, while processing of an adequate number of in vivo virtual biopsy images provides quantitative and objective bio-information.

Subpixel interpolation is an essential part to increase coding efficiency in many video compression standards. There arises the need to support the interpolation processes in multiple standards. In this paper, we propose a reconfigurable interpolation architecture for video decoding in MPEG-2, MPEG-4, and AVC/H.264. To reduce the hardware complexity and improve the operation efficiency, we analyze the commonality of interpolation filters for target standards. To have efficient memory access, we present a method to arrange data properly in cache memory. Our interpolator adopts the high throughput separated 1-D design and can be adapted to each target standard by using reconfigurable interpolation filters. We design the reconfigurable interpolation filter based on the commonality analysis so that it has very simple structure. The implementation results show that our interpolator consuming 31.7K gates can support processing video with resolution of 1920?×?1088 and frame rate of 30 frames/s. This compares very favorably with the existing architectures in silicon area and performance.

This paper presents a reconfigurable architecture for two video coding standards, MPEG-2 and AVC/H.264. The proposed reconfigurable architecture dynamically configures the architecture to achieve multiple functions in variable length coding decoder and context-adaptive variable length coding decoder. By means of extracting the commonalities at the low-level dataflow, the designed reconfigurable architecture optimizes hardware resources and increases architectural flexibility through the reconfigurability in the adaptively constructing architecture. On the other hand, the proposed reconfigurable architecture combines several strategies in AVC/H.264 mode, and subsequently revises and integrates these mechanisms to achieve a high-throughput rate. As a consequence, the memory usage of the proposed reconfigurable architecture achieves 37.5 % memory bits saving and 46.7 % memory area reduction as compared to the individual implementation; moreover, in AVC/H.264 mode, this reconfigurable architecture requires only 50 cycles to decode one marcoblock on average. Due to the flexibility, the proposed reconfigurable architecture costs 10.8 K gates by using 0.18 μm CMOS technology at the 108 MHz clock rate, and supports variable length coding in MPEG-2 with 1920?×?1080@30fps and context-adaptive variable length coding in AVC/H.264 with 1920?×?1080@60fps. From the perspectives of architectural costs and supported functions, the proposed architecture surpasses the related works in the state of the art.

This paper presents an algorithm that is capable of generating 3-D video from monocular video through a hierarchical approach that characterizes the low-level features and high-level semantics of video content to generate depth map. Color and texture features are used to locally indicate video's characteristics followed by segmentation of the video into several regions. Subsequently, semantics, including shape and motion semantics, of segmented regions are delineated from a higher perspective to refine segmentation result via measuring the interrelations among segmented regions. Subsequently, according to refined segmentation map and semantics of segmented regions, the proposed method generates a depth map with high stability using both spatial and temporal information. The stable depth map minimized visual quality degradation, such as flicker and blurring when viewing the 3-D video. The experimental results show the capability of the proposed algorithm in generating a high quality and stable depth map. In addition, subjective viewing evaluation shows that the proposed algorithm surpasses those of the commercial products for converting 2-D video to 3-D video, including TriDef 3D and CyberLink Power DVD.

This paper presents bi-directional trajectory tracking with variable block-size motion estimation for frame rate up-conversion (FRUC) based on the algorithm/architecture co-exploration (AAC) design methodology. Due to concurrent exploration in both algorithm and architecture domains, the designed system requires fewer computations and lowers hardware cost, but is capable of enhancing the accuracy of motion vectors (MVs) by allowing MV refinement from coarse-grained to fine-grained. In addition, a method that uses multiple block candidates tracked by bi-directional MVs for interpolation is also presented to improve visual quality. Benefiting from AAC, we can extract architectural information at algorithmic design phase to determine the most feasible architecture; then, the proposed algorithm can be mapped onto target platform smoothly. The proposed FRUC system, which is capable of converting the frame rate from 60 fps up to 120 fps at full HD (1920 × 1080) resolution, was successfully implemented and verified on a field-programmable gate array. This FRUC system's performance has not only been shown to surpass state-of-art alternatives in algorithmic performance, but its hardware cost is less than the comparable works described in the literature.

Traditional biopsy procedures require invasive tissue removal from a living subject, followed by time-consuming and complicated processes, so noninvasive in vivo virtual biopsy, which possesses the ability to obtain exhaustive tissue images without removing tissues, is highly desired. Some sets of in vivo virtual biopsy images provided by healthy volunteers were processed by the proposed cell segmentation approach, which is based on the watershed-based approach and the concept of convergence index filter for automatic cell segmentation. Experimental results suggest that the proposed algorithm not only reveals high accuracy for cell segmentation but also has dramatic potential for noninvasive analysis of cell nuclear-to-cytoplasmic ratio (NC ratio), which is important in identifying or detecting early symptoms of diseases with abnormal NC ratios, such as skin cancers during clinical diagnosis via medical imaging analysis.

Deep Convolutional Neural Network (CNN) draws significant attention in the computer vision community by facilitating machines with more intelligence in understanding visual signals, however, its computation complexity has also increased significantly. To achieve ubiquitous machine intelligence, deep CNN is required to be ported onto local devices rather than cloud-based solution due to low latency consideration. Hence, in this paper, we propose a method to explore the design space for porting deep CNN onto iOS mobile devices, with attempts in maximizing data reusability, which alleviates the high bandwidth burden in the convolution layers of CNN. Furthermore, effective data reuse also makes possible the parallelization of all computing threads without data loading latency. On the other hand, deep CNN is usually over-parametrized with many unused convolution kernels. Based on Algorithm/Architecture Co-Exploration, we introduced a method in pruning redundant kernels in deep CNN with ignorable performance degradation on validation dataset (0.06% loss). This reduces 29% of operations and 34% of storage size on a 16-layer CNN. We used iPhone 6s and iPad Pro for case studies, and ported 8-layer and 16-layer CNNs onto targeted devices. The data reusability strategy improves computation speed up to 1.3x, and redundant kernel removal increases computation speed to 1.43x. As a result, we achieved high computation efficiency and have thus enhanced the capability of machine intelligence on local mobile devices.

Concurrently exploring both algorithmic and architectural optimizations is a new design paradigm. This survey paper addresses the latest research and future perspectives on the simultaneous development of video coding, processing, and computing algorithms with emerging platforms that have multiple cores and reconfigurable architecture. As the algorithms in forthcoming visual systems become increasingly complex, many applications must have different profiles with different levels of performance. Hence, with expectations that the visual experience in the future will become continuously better, it is critical that advanced platforms provide higher performance, better flexibility, and lower power consumption. To achieve these goals, algorithm and architecture co-design is significant for characterizing the algorithmic complexity used to optimize targeted architecture. This paper shows that seamless weaving of the development of previously autonomous visual computing algorithms and multicore or reconfigurable architectures will unavoidably become the leading trend in the future of video technology.

This paper presents the algorithm and reconfigurable architecture of motion-adaptive deinterlacer for high-definition video. The content-adaptability of algorithm and the reconfiguration of architecture are concurrently explored by algorithm/architecture co-design methodology and Caltrop actor language (CAL) modeling of the dataflow. In the design methodology we employed, the CAL dataflow model is also very helpful in the verification of our deinerlacer. The proposed algorithm and architecture design of deinterlacer is more cost-efficient than two recently proposed works in terms of algorithmic performance and silicon area of VLSI implementation. Moreover, data path reconfiguration efficiently enables various interpolation schemes using less computational resource of hardware than non-reconfigurable architecture.

An efficient rate control algorithm based on the content-adaptive initial quantisation parameter (QP) setting scheme and the peak signal-to-noise ratio (PSNR) variation-limited bit-allocation strategy for low-complexity mobile applications is presented. This algorithm can efficiently measure the residual complexity of intra-pictures without performing the computation-intensive intra-prediction and mode decision in H.264/AVC, based on the structural and statistical features of local textures. This can adaptively set proper initial QP values for versatile video contents. In addition, this bit-allocation strategy can effectively distribute bit-rate budgets based on the monotonic property to enhance overall coding efficiency while maintaining the consistency of visual quality by limiting the variation of quantisation distortion. The experimental results reveal that the proposed algorithm surpasses the conventional rate control approaches in terms of the average PSNR from 0.34 to 0.95 dB. Moreover, this algorithm provides more impressive visual quality and more robust buffer controllability when compared with other algorithms.

This paper introduces a texture analysis mechanism utilizing multiresolution technique to reduce false motion detection and hence thoroughly improve the interpolation results for high-quality deinterlacing. Conventional motion-adaptive deinterlacing algorithm selects from inter-field and intra-field interpolations according to motion. Accurate determination of motion information is essential for this purpose. Fine textures, having high local pixel variation, tend to cause false detection of motion. Based on hierarchical wavelet analysis, this algorithm provides much better perceptual visual quality and considerably higher PSNR than other motion adaptive deinterlacers as shown. In addition, a recursive 3-field motion detection algorithm is also proposed to achieve better performance than the traditional 2-field motion detection algorithm with little memory overhead.

This paper presents a new spatio-temporal motion estimation algorithm and its VLSI architecture for video coding based on algorithm and architecture co-design methodology. The algorithm consists of the new strategies of spatio-temporal motion vector prediction, modified one-at-a-time search scheme, and multiple update paths derived from optimization theory. The hardware specification is for high-definition video coding. We applied the ME algorithm to H.264 reference software. Our algorithm surpasses recently published research and achieves close performance to full search. The VLSI implementation proves the low cost feature of our algorithm. The algorithm and architecture co-design concept is highly emphasized in this paper. We provide some quantitative example to show the necessity of algorithm and architecture co-design

Reference EPFL-ARTICLE-159226doi:10.1109/TCSVT.2009.2034438View record in Web of Science Record created on 2010-11-30, modified on 2017-05-10

In this paper a multiresolution wavelet analysis (MWA) and nonstationary Gaussian Markov random field (GMRF) technique is introduced for the detection of microcalcifications with high accuracy. The hierarchical multiresolution wavelet information in conjunction with the contextual information of the images extracted from GMRF provides an efficient technique for microcalcification detection. A Bayesian learning paradigm realized via the expectation maximization (EM) algorithm was also introduced for edge detection or segmentation of mass regions recorded on the mammograms. The strength of the technique is in the effective utilization of the rich contextural information in the images considered. The effectiveness of the approach has been tested with a number of mammographic images for which the microcalcification detection algorithm achieved a sensitivity (true positive rate) of 94% and specificity (true negative rate) of 88%. Considerably good results were also obtained for the segmentation algorithm. In addition, the results for both the detected microcalcifications and the segmented mass regions were superimposed for an interesting case under the methods introduced.

This article presents a novel algorithm for image seg- been developed for classification purposes. In addition, many mentation via the use of the multiresolution wavelet analysis and the authors have discovered significant advantages in the use of the expectation maximization (EM) algorithm. The development of a multiresolution concept ( 4,5 ) . Brazkovic and Neskovic presented multiresolution wavelet feature extraction scheme is based on the the Gaussian pyramid and fuzzy linking method for the adaptive Gaussian Markov random field (GMRF) assumption in mammo- detection of cancerous changes in mammograms (6). graphic image modeling. Mammographic images are hierarchically Recently, as a result of cross-fertilization of innovative ideas decomposed into different resolutions. In general, larger breast le- from image processing, spatial statistics, and statistical physics, sions are characterized by coarser resolutions, whereas higher resolu- a significant amount of research activity on image modeling and tions show finer and more detailed anatomical structures. These hier- archical variations in the anatomical features displayed by multiresolu- segmentation has also been concentrated on the two-dimensional tion decomposition are further quantified through the application of ( 2D ) Markov random field ( MRF ) . Although many of the poten- the Gaussian Markov random field. Because of its uniqueness in local- tials of MRF had been envisioned by the early works of Levy ity, adaptive features based on the nonstationary assumption of (7), McCormick and Jayaramamrhy (8), and Abend et al. (9), GMRF are defined for each pixel of the mammogram. Fibroadenomas exploitation of the powers of the MRF was not possible until are then segmented via the fuzzy C-means algorithm using these significant recent advances occurred in the appropriate mathemat- localized features. Subsequently, the segmentation results are further ical and computational tools. Chellappa and Kashyap (10 ) suc- enhanced via the introduction of a maximum a posteriori (MAP) seg- cessfully applied the noncausal autoregressive ( NCAR ) model mentation estimation scheme based on the Bayesian learning para