No abstract

In order to meet the requirement of high sensitivity and signal-to-noise ratios (SNR), this study develops and optimizes a piezoresistive pressure sensor by using double silicon nanowire (SiNW) as the piezoresistive sensing element. First of all, ANSYS finite element method and voltage noise models are adopted to optimize the sensor size and the sensor output (such as sensitivity, voltage noise and SNR). As a result, the sensor of the released double SiNW has 1.2 times more sensitivity than that of single SiNW sensor, which is consistent with the experimental result. Our result also displays that both the sensitivity and SNR are closely related to the geometry parameters of SiNW and its doping concentration. To achieve high performance, a p-type implantation of 5 × 1018 cm−3 and geometry of 10 µm long SiNW piezoresistor of 1400 nm × 100 nm cross area and 6 µm thick diaphragm of 200 µm × 200 µm are required. Then, the proposed SiNW pressure sensor is fabricated by using the standard complementary metal-oxide-semiconductor (CMOS) lithography process as well as wet-etch release process. This SiNW pressure sensor produces a change in the voltage output when the external pressure is applied. The involved experimental results show that the pressure sensor has a high sensitivity of 495 mV/V·MPa in the range of 0–100 kPa. Nevertheless, the performance of the pressure sensor is influenced by the temperature drift. Finally, for the sake of obtaining accurate and complete information over wide temperature and pressure ranges, the data fusion technique is proposed based on the back-propagation (BP) neural network, which is improved by the particle swarm optimization (PSO) algorithm. The particle swarm optimization–back-propagation (PSO–BP) model is implemented in hardware using a 32-bit STMicroelectronics (STM32) microcontroller. The results of calibration and test experiments clearly prove that the PSO–BP neural network can be effectively applied to minimize sensor errors derived from temperature drift.

Human-motion generation is a long-standing challenging task due to the requirement of accurately modeling complex and diverse dynamic patterns. Most existing methods adopt sequence models such as RNN to directly model transitions in the original action space. Due to high dimensionality and potential noise, such modeling of action transitions is particularly challenging. In this paper, we focus on skeleton-based action generation and propose to model smooth and diverse transitions on a latent space of action sequences with much lower dimensionality. Conditioned on a latent sequence, actions are generated by a frame-wise decoder shared by all latent action-poses. Specifically, an implicit RNN is defined to model smooth latent sequences, whose randomness (diversity) is controlled by noise from the input. Different from standard action-prediction methods, our model can generate action sequences from pure noise without any conditional action poses. Remarkably, it can also generate unseen actions from mixed classes during training. Our model is learned with a bi-directional generative-adversarial-net framework, which can not only generate diverse action sequences of a particular class or mix classes, but also learns to classify action sequences within the same model. Experimental results show the superiority of our method in both diverse action-sequence generation and classification, relative to existing methods.

No abstract

Fumonisin B(1) (FB(1)) is the most abundant of the fumonisin mycotoxins, mainly produced in maize by F. verticillioides and F. proliferatum. A total of 282 corn samples harvested in 2005 from six provinces, the main corn-producing areas of China, were analyzed for FB(1) using high-performance liquid chromatography. All samples except one were (99.6%) positive for FB(1) at levels varying from 3 to 71,121 ng/g with mean and median levels for all samples of 6,662 and 1,569 ng/g, respectively. During an analysis of the distribution pattern for FB(1), it became apparent that 43.6% of tested samples had FB(1) concentrations below 1,000 ng/g, while 25.2% contained in excess of 5,000 ng/g. The average exposure to FB(1) (1.1 microg/kg body weight/day) is within the provisional maximum tolerable daily intake of 2 microg/kg body weight/day set by the Joint FAO/WHO Expert Committee on Food Additives.

Human-motion generation is a long-standing challenging task due to the requirement of accurately modeling complex and diverse dynamic patterns. Most existing methods adopt sequence models such as RNN to directly model transitions in the original action space. Due to high dimensionality and potential noise, such modeling of action transitions is particularly challenging. In this paper, we focus on skeleton-based action generation and propose to model smooth and diverse transitions on a latent space of action sequences with much lower dimensionality. Conditioned on a latent sequence, actions are generated by a frame-wise decoder shared by all latent action-poses. Specifically, an implicit RNN is defined to model smooth latent sequences, whose randomness (diversity) is controlled by noise from the input. Different from standard action-prediction methods, our model can generate action sequences from pure noise without any conditional action poses. Remarkably, it can also generate unseen actions from mixed classes during training. Our model is learned with a bi-directional generative-adversarial-net framework, which not only can generate diverse action sequences of a particular class or mix classes, but also learns to classify action sequences within the same model. Experimental results show the superiority of our method in both diverse action-sequence generation and classification, relative to existing methods.

In this paper, we present a spectral intrinsic image decomposition (SIID) model, which is dedicated to resolve a natural scene into its purely independent intrinsic components: illumination, shading, and reflectance. By introducing spectral information, our work can solve many challenging cases, such as scenes with metameric effects, which are hard to tackle for trichromatic intrinsic image decomposition (IID), and thus offers potential benefits to many higher-level vision tasks, e.g., materials classification and recognition, shape-from-shading, and spectral image relighting. A both effective and efficient algorithm is presented to decompose a spectral image into its independent intrinsic components. To facilitate future SIID research, we present a public dataset with ground-truth illumination, shading, reflectance and specularity, and a meaningful error metric, so that the quantitative comparison becomes achievable. The experiments on this dataset and other images demonstrate the accuracy and robustness of the proposed method on diverse scenes, and reveal that more spectral channels indeed facilitate the vision task (i.e., segmentation and recognition).

In fluorescence microscopy, computational algorithms have been developed to suppress noise, enhance contrast, and even enable super-resolution (SR). However, the local quality of the images may vary on multiple scales, and these differences can lead to misconceptions, which is especially intractable in emerging deep-learning ones. Current mapping methods fail to finely estimate the local quality, challenging to associate the SR scale content. Here, we develop a rolling Fourier ring correlation (rFRC) framework to evaluate the reconstruction uncertainties down to SR scale. To visually pinpoint regions with low reliability, a filtered rFRC is combined with a modified resolution scaled error map (RSM), offering a comprehensive and concise map for further examination. We demonstrate their performances on various SR imaging modalities, and the resulting quantitative maps enable better SR images integrated from different reconstructions. Beyond that, we provide a strategy for learning-based restorations, allowing a direct detection of both data and model uncertainties, and expect the representative cases can inspire further advances in this rapidly developing field.