After research on the motion estimation algorithm in video coding, a motion estimator algorithm with 1/4 pixel ac- curacy is implemented based on fie/d-programmable gate array （FPGA）. The motion estimation algorithm...After research on the motion estimation algorithm in video coding, a motion estimator algorithm with 1/4 pixel ac- curacy is implemented based on fie/d-programmable gate array （FPGA）. The motion estimation algorithm module is made up of the 1[4 pixel interpolation module with serial input and parallel output, the three step search module and the block match- ing module, which can use relatively less Wiener filters for interpolation operation. Experiment results show that the hard- ware design has less consumption of the logical resource, higher stability and lower power consumption.展开更多
In this paper, we study how pixel size influences energy resolution for a proposed pixelated detector—a high sensitivity, low cost, and real-time radon monitor based on a Topmetal-Ⅱ^- time projection chamber(TPC). T...In this paper, we study how pixel size influences energy resolution for a proposed pixelated detector—a high sensitivity, low cost, and real-time radon monitor based on a Topmetal-Ⅱ^- time projection chamber(TPC). This monitor was designed to improve spatial resolution for detecting radon alpha particles using Topmetal-Ⅱ^- sensors assembled by a 0.35 lm CMOS integrated circuit process.Owing to concerns that small pixel size might have the side effect of worsening energy resolution due to lower signalto-noise ratio, a Geant4-based simulation was used to investigate the dependence of energy resolution on pixel sizes ranging from 60 to 600 lm. A non-monotonic trend in this region shows the combined effect of pixel size and threshold on pixels, analyzed by introducing an empirical expression. Pixel noise contributes 50 keV full-width at half-maximum energy resolution for 400 lm pixel size at 1–4σ threshold that is comparable to the energy resolution caused by energy fluctuations in the TPC ionization process( ~20 keV). The total energy resolution after combining both factors is estimated to be 54 keV for a pixel size of 400 lm at 1–4σ threshold. The analysis presented in this paper would help choosing suitable pixel size for future pixelated detectors.展开更多
By using the MOS-based model established in this paper, the physical process of photoelectron generation, transfer,and storage in the four-transistor active pixel sensor(4 T-APS) pixels can be simulated in SPICE envir...By using the MOS-based model established in this paper, the physical process of photoelectron generation, transfer,and storage in the four-transistor active pixel sensor(4 T-APS) pixels can be simulated in SPICE environment. The variable capacitance characteristics of double junctions in pinned photodiodes(PPDs) and the threshold voltage difference formed by channel nonuniform doping in transfer gates(TGs) are considered with this model. The charge transfer process of photogenerated electrons from PPDs to the floating diffusion(FD) is analyzed, and the function of nonuniform doping of TGs in suppressing charge injection back to PPDs is represented with the model. The optical and electrical characteristics of all devices in the pixel are effectively combined with the model. Moreover, the charge transfer efficiency and the voltage variation in PPD can be described with the model. Compared with the hybrid simulation in TCAD and the Verilog-A simulation in SPICE, this model has higher simulation efficiency and accuracy, respectively. The effectiveness of the MOS-based model is experimentally verified in a 3 μm test pixel designed in 0.11 μm CIS process.展开更多
在视频编码系统中,半像素精度的运动估计虽可以明显地改善编码效果,但也因此增加了不少运算量。为了降低运算量和提高半像素运动估计的速度,提出了一种全新的半像素搜索算法(half-pixel motion estimation based on linear prediction,...在视频编码系统中,半像素精度的运动估计虽可以明显地改善编码效果,但也因此增加了不少运算量。为了降低运算量和提高半像素运动估计的速度,提出了一种全新的半像素搜索算法(half-pixel motion estimation based on linear prediction,简称BLPHME),其关键思想是通过分析整像素搜索和半像素搜索结果之间的相关性来建立一个线性模型,通过动态调整判决门限,以便预测并跳过那些不能从半像素块匹配搜索中得到好处的块。实验结果表明,该算法不仅可以明显地降低运动估计的运算量,同时还能得到与传统算法非常接近的图像质量和码率。此外,该算法还可以和基于整像素和半像素的快速运动估计算法一起使用,以进一步降低运算量。展开更多
基金The 8th Graduate Technology Fund of North University of China(No.201111)
文摘After research on the motion estimation algorithm in video coding, a motion estimator algorithm with 1/4 pixel ac- curacy is implemented based on fie/d-programmable gate array （FPGA）. The motion estimation algorithm module is made up of the 1[4 pixel interpolation module with serial input and parallel output, the three step search module and the block match- ing module, which can use relatively less Wiener filters for interpolation operation. Experiment results show that the hard- ware design has less consumption of the logical resource, higher stability and lower power consumption.
基金supported by the National Natural Science Foundation of China(No.U1732271)
文摘In this paper, we study how pixel size influences energy resolution for a proposed pixelated detector—a high sensitivity, low cost, and real-time radon monitor based on a Topmetal-Ⅱ^- time projection chamber(TPC). This monitor was designed to improve spatial resolution for detecting radon alpha particles using Topmetal-Ⅱ^- sensors assembled by a 0.35 lm CMOS integrated circuit process.Owing to concerns that small pixel size might have the side effect of worsening energy resolution due to lower signalto-noise ratio, a Geant4-based simulation was used to investigate the dependence of energy resolution on pixel sizes ranging from 60 to 600 lm. A non-monotonic trend in this region shows the combined effect of pixel size and threshold on pixels, analyzed by introducing an empirical expression. Pixel noise contributes 50 keV full-width at half-maximum energy resolution for 400 lm pixel size at 1–4σ threshold that is comparable to the energy resolution caused by energy fluctuations in the TPC ionization process( ~20 keV). The total energy resolution after combining both factors is estimated to be 54 keV for a pixel size of 400 lm at 1–4σ threshold. The analysis presented in this paper would help choosing suitable pixel size for future pixelated detectors.
基金Project supported by the National Natural Science Foundation of China (Grant No. 61874085)the Postdoctoral Research Funding Project of Shaanxi Province,China (Grant No. 2018BSHEDZZ41)。
文摘By using the MOS-based model established in this paper, the physical process of photoelectron generation, transfer,and storage in the four-transistor active pixel sensor(4 T-APS) pixels can be simulated in SPICE environment. The variable capacitance characteristics of double junctions in pinned photodiodes(PPDs) and the threshold voltage difference formed by channel nonuniform doping in transfer gates(TGs) are considered with this model. The charge transfer process of photogenerated electrons from PPDs to the floating diffusion(FD) is analyzed, and the function of nonuniform doping of TGs in suppressing charge injection back to PPDs is represented with the model. The optical and electrical characteristics of all devices in the pixel are effectively combined with the model. Moreover, the charge transfer efficiency and the voltage variation in PPD can be described with the model. Compared with the hybrid simulation in TCAD and the Verilog-A simulation in SPICE, this model has higher simulation efficiency and accuracy, respectively. The effectiveness of the MOS-based model is experimentally verified in a 3 μm test pixel designed in 0.11 μm CIS process.
文摘在视频编码系统中,半像素精度的运动估计虽可以明显地改善编码效果,但也因此增加了不少运算量。为了降低运算量和提高半像素运动估计的速度,提出了一种全新的半像素搜索算法(half-pixel motion estimation based on linear prediction,简称BLPHME),其关键思想是通过分析整像素搜索和半像素搜索结果之间的相关性来建立一个线性模型,通过动态调整判决门限,以便预测并跳过那些不能从半像素块匹配搜索中得到好处的块。实验结果表明,该算法不仅可以明显地降低运动估计的运算量,同时还能得到与传统算法非常接近的图像质量和码率。此外,该算法还可以和基于整像素和半像素的快速运动估计算法一起使用,以进一步降低运算量。
