No abstract

To realize next-generation highly realistic sensation broadcasting systems, the research and development of 8K ultrahigh-definition television (UHDTV) systems have been promoted. To realize 8K video cameras, 33Mpixel sensors [1-2] and a full-resolution camera system that uses three 33Mpixel sensors [3] have been reported. However, the weight of the camera with three sensors is over 40kg because the camera requires a large-format color-separation prism. To reduce the size of the camera, single-chip imaging is a promising approach, and a compact single-chip 8K camera that weighs only 2kg has already been developed using a color 33Mpixel CMOS image sensor [4]. However, a conventional single-chip camera has a lower image quality than a full-resolution camera because the total pixel count of the single-sensor camera is only one-third of that of the three-sensor camera, and pixel interpolation is required to configure a full-resolution image. In this paper, a 133Mpixel sensor that can be operated at 60fps to realize a full-resolution 8K single-chip camera is described. To achieve both high speed and suitable ADC resolution, 32-column multiplexing analog readout circuitry and 14b high-speed redundant successive approximation register (SAR) ADCs [5] are adopted. As a result, a full-size image with a data rate of 128.71Gb/s at 60fps has been captured. Figure 6.2.1 shows the sensor block diagram. The total pixel array size is 15488(H) × 8776(V) including optical black pixels. The pixel design is a 2.45μm two-way vertically shared pinned photodiode. The readout of the array is addressed one physical pixel row at a time with 15,488 readout columns split between the top and bottom of the pixel array. The column readout circuit consists of a source-follower bias current (VLN), a programmable-gain amplifier (PGA), and two sets of sample-and-hold capacitor banks (SHCaps) for even-and-odd-row ping-pong operation. Analog data stored in the SHCaps are read out in parallel by 484 SAR-ADCs. Each SAR-ADC serially multiplexes 32 columns, and the resulting converter data are written into the SRAM line memory. Two banks of SRAM memory are used to provide pipelining of the ADC and SRAM readout to reduce the row time. The SRAM readout is further divided into 16 parallel ports to reduce the data rate. Each readout port outputs 960 columns, with the exception of the 4 edge corner ports that output an additional 32 optical black columns. CML output drivers are used to output 7b-wide data to a 574.56MHz DDR, achieving an aggregate data rate of 128.71Gb/s. The sensor block diagram is completed with distributed PLLs, SPI communication, and on-chip timing control logic. Figure 6.2.2 shows a simplified block diagram of the signal readout path and timing control logic blocks. The PGA, SHCaps, CMR, and ADC are similar to those described in [5], which enables us to design a large-format sensor at 60fps that fits into a 1D stitching reticle. Front-end multiplexing of the columns is optimized for 32 columns per CMR and a 14b SAR-ADC with a conversion r...

We propose a geometric convexity shape prior preservation method for variational level set based image segmentation methods. Our method is built upon the fact that the level set of a convex signed distanced function must be convex. This property enables us to transfer a complicated geometrical convexity prior into a simple inequality constraint on the function. An active set based Gauss-Seidel iteration is used to handle this constrained minimization problem to get an efficient algorithm. We apply our method to region and edge based level set segmentation models including Chan-Vese (CV) model with guarantee that the segmented region will be convex. Experimental results show the effectiveness and quality of the proposed model and algorithm.

Article is available under the Creative Commons Attribution-Noncommercial-NoDerivs 3.0 license CC BY-NC-ND 3.0.

No abstract

In order to study the compatibility between polycarboxyIn order to study the compatibility between polycarboxylates superplasticizer (PC) and different kinds of retarders, sodium citrate (SC), sodium gluconate (SG), sodium hexametaphosphate (SH) and sucrose (SU) were compounded with PC respectively in the same proportion. The liquid degree and its loss of cement paste, the setting time of cement, the blending strength and compressive strength of mortar that admixed with different kinds of retarders were measured, and the influence mechanism of the compatibility between different kinds of retarders and PC were analyzed. The experiment results showed that it was benefit to the liquid degree of cement paste after SG or SH or SU was compounded with PC; The setting time of cement could prolonged more by SH or SU; Blending and compressive strength of mortar in different ages were enhanced in large extent after SG were adopted; The strength of mortar was reduced in early age and enhanced in small extent in mid-later age after SU was compounded. Comprehensively considering to the influence of liquid degree, setting time and strength, the compatibility of SG with PC is higher than other kinds of retarders.

A dual expectation-maximization (EM) algorithm for total variation (TV) regularized Gaussian mixture model (GMM) is proposed in this paper. The algorithm is built upon the EM algorithm with TV regularization (EM-TV) model which combines the statistical and variational methods together for image segmentation. Inspired by the projection algorithm proposed by Chambolle, we give a dual algorithm for the EM-TV model. The related dual problem is smooth and can be easily solved by a projection gradient method, which is stable and fast. Given the parameters of GMM, the proposed algorithm can be seen as a forward-backward splitting method which converges. This method can be easily extended to many other applications. Numerical results show that our algorithm can provide high quality segmentation results with fast computation speed. Compared with the well-known statistics based methods such as hidden Markov random field with EM method (HMRF-EM), the proposed algorithm has a better performance. The proposed method could also be applied to MRI segmentation such as SPM8 software and improve the segmentation results.

No abstract