ALBERT

Description: Heavy occlusions in cluttered scenes impose significant challenges to many computer vision applications. Recent light field imaging systems provide new see-through capabilities through synthetic aperture imaging (SAI) to overcome the occlusion problem. Existing synthetic aperture imaging methods, however, emulate focusing at a specific depth layer, but are incapable of producing an all-in-focus see-through image. Alternative in-painting algorithms can generate visually-plausible results, but cannot guarantee the correctness of the results. In this paper, we present a novel depth-free all-in-focus SAI technique based on light field visibility analysis. Specifically, we partition the scene into multiple visibility layers to directly deal with layer-wise occlusion and apply an optimization framework to propagate the visibility information between multiple layers. On each layer, visibility and optimal focus depth estimation is formulated as a multiple-label energy minimization problem. The layer-wise energy integrates all of the visibility masks from its previous layers, multi-view intensity consistency and depth smoothness constraint together. We compare our method with state-of-the-art solutions, and extensive experimental results demonstrate the effectiveness and superiority of our approach.

Description: For real-time solution of inertial navigation system (INS), the high-degree spherical harmonic gravity model (SHM) is not applicable because of its time and space complexity, in which traditional normal gravity model (NGM) has been the dominant technique for gravity compensation. In this paper, a two-dimensional second-order polynomial model is derived from SHM according to the approximate linear characteristic of regional disturbing potential. Firstly, deflections of vertical (DOVs) on dense grids are calculated with SHM in an external computer. And then, the polynomial coefficients are obtained using these DOVs. To achieve global navigation, the coefficients and applicable region of polynomial model are both updated synchronously in above computer. Compared with high-degree SHM, the polynomial model takes less storage and computational time at the expense of minor precision. Meanwhile, the model is more accurate than NGM. Finally, numerical test and INS experiment show that the proposed method outperforms traditional gravity models applied for high precision free-INS.

Description: Sensors, Vol. 18, Pages 1530: Convolutional Neural Network-Based Embarrassing Situation Detection under Camera for Social Robot in Smart Homes Sensors doi: 10.3390/s18051530 Authors: Guanci Yang Jing Yang Weihua Sheng Francisco Erivaldo Fernandes Junior Shaobo Li Recent research has shown that the ubiquitous use of cameras and voice monitoring equipment in a home environment can raise privacy concerns and affect human mental health. This can be a major obstacle to the deployment of smart home systems for elderly or disabled care. This study uses a social robot to detect embarrassing situations. Firstly, we designed an improved neural network structure based on the You Only Look Once (YOLO) model to obtain feature information. By focusing on reducing area redundancy and computation time, we proposed a bounding-box merging algorithm based on region proposal networks (B-RPN), to merge the areas that have similar features and determine the borders of the bounding box. Thereafter, we designed a feature extraction algorithm based on our improved YOLO and B-RPN, called F-YOLO, for our training datasets, and then proposed a real-time object detection algorithm based on F-YOLO (RODA-FY). We implemented RODA-FY and compared models on our MAT social robot. Secondly, we considered six types of situations in smart homes, and developed training and validation datasets, containing 2580 and 360 images, respectively. Meanwhile, we designed three types of experiments with four types of test datasets composed of 960 sample images. Thirdly, we analyzed how a different number of training iterations affects our prediction estimation, and then we explored the relationship between recognition accuracy and learning rates. Our results show that our proposed privacy detection system can recognize designed situations in the smart home with an acceptable recognition accuracy of 94.48%. Finally, we compared the results among RODA-FY, Inception V3, and YOLO, which indicate that our proposed RODA-FY outperforms the other comparison models in recognition accuracy.

Description: Sensors, Vol. 18, Pages 1552: A Combined Gravity Compensation Method for INS Using the Simplified Gravity Model and Gravity Database Sensors doi: 10.3390/s18051552 Authors: Xiao Zhou Gongliu Yang Jing Wang Zeyang Wen In recent decades, gravity compensation has become an important way to reduce the position error of an inertial navigation system (INS), especially for a high-precision INS, because of the extensive application of high precision inertial sensors (accelerometers and gyros). This paper first deducts the INS’s solution error considering gravity disturbance and simulates the results. Meanwhile, this paper proposes a combined gravity compensation method using a simplified gravity model and gravity database. This new combined method consists of two steps all together. Step 1 subtracts the normal gravity using a simplified gravity model. Step 2 first obtains the gravity disturbance on the trajectory of the carrier with the help of ELM training based on the measured gravity data (provided by Institute of Geodesy and Geophysics; Chinese Academy of sciences), and then compensates it into the error equations of the INS, considering the gravity disturbance, to further improve the navigation accuracy. The effectiveness and feasibility of this new gravity compensation method for the INS are verified through vehicle tests in two different regions; one is in flat terrain with mild gravity variation and the other is in complex terrain with fierce gravity variation. During 2 h vehicle tests, the positioning accuracy of two tests can improve by 20% and 38% respectively, after the gravity is compensated by the proposed method.

Description: Sensors, Vol. 18, Pages 2312: Amplitude-Based Filtering for Video Magnification in Presence of Large Motion Sensors doi: 10.3390/s18072312 Authors: Xiu Wu Xuezhi Yang Jing Jin Zhao Yang Video magnification reveals important and informative subtle variations in the world. These signals are often combined with large motions which result in significant blurring artifacts and haloes when conventional video magnification approaches are used. To counter these issues, this paper presents an amplitude-based filtering algorithm that can magnify small changes in video in presence of large motions. We seek to understand the amplitude characteristic of small changes and large motions with the goal of extracting accurate signals for visualization. Based on spectrum amplitude filtering, the large motions can be removed while small changes can still be magnified by Eulerian approach. An advantage of this algorithm is that it can handle large motions, whether they are linear or nonlinear. Our experimental results show that the proposed method can amplify subtle variations in the presence of large motion, as well as significantly reduce artifacts. We demonstrate the presented algorithm by comparing to the state-of-the-art and provide subjective and objective evidence for the proposed method.

Description: A self-sustained multi-sensor platform for indoor environmental monitoring is proposed in this paper. To reduce the cost and power consumption of the sensing platform, in the developed platform, organic materials of PEDOT:PSS and PEDOT:PSS/EB-PANI are used as the sensing films for humidity and CO2 detection, respectively. Different from traditional gas sensors, these organic sensing films can operate at room temperature without heating processes or infrared transceivers so that the power consumption of the developed humidity and the CO2 sensors can be as low as 10 μW and 5 μW, respectively. To cooperate with these low-power sensors, a Complementary Metal-Oxide-Semiconductor (CMOS) system-on-chip (SoC) is designed to amplify and to read out multiple sensor signals with low power consumption. The developed SoC includes an analog-front-end interface circuit (AFE), an analog-to-digital convertor (ADC), a digital controller and a power management unit (PMU). Scheduled by the digital controller, the sensing circuits are power gated with a small duty-cycle to reduce the average power consumption to 3.2 μW. The designed PMU converts the power scavenged from a dye sensitized solar cell (DSSC) module into required supply voltages for SoC circuits operation under typical indoor illuminance conditions. To our knowledge, this is the first multiple environmental parameters (Temperature/CO2/Humidity) sensing platform that demonstrates a true self-powering functionality for long-term operations.

Description: The goal of crowd counting is to estimate the number of people in the image. Presently, use regression to count people number became a mainstream method. It is worth noting that, with the development of convolutional neural networks (CNN), methods that are based on CNN have become a research hotspot. It is a more interesting topic that how to locate the site of the person in the image than simply predicting the number of people in the image. The perspective transformation present is still a challenge, because perspective distortion will cause differences in the size of the crowd in the image. To devote perspective distortion and locate the site of the person more accuracy, we design a novel framework named Adaptive Learning Network (CAL). We use the VGG as the backbone. After each pooling layer is output, we collect the 1/2, 1/4, 1/8, and 1/16 features of the original image and combine them with the weights learned by an adaptive learning branch. The object of our adaptive learning branch is each image in the datasets. By combining the output features of different sizes of each image, the challenge of drastic changes in the size of the image crowd due to perspective transformation is reduced. We conducted experiments on four population counting data sets (i.e., ShanghaiTech Part A, ShanghaiTech Part B, UCF_CC_50 and UCF-QNRF), and the results show that our model has a good performance.

Description: Aluminum is the most abundant metallic element in the Earth’s crust and acts as a non-essential element for biological species. The accumulation of excessive amounts of aluminum can be harmful to biological species. Thus, the development of convenient and selective tools for the aluminum detection is necessary. In this work, a highly selective aluminum ion fluorescent probe N’-(2,5-dihydroxybenzylidene)acetohydrazide (Al-II) has been successfully synthesized and systemically characterized. The fluorescence intensity of this probe shows a significant enhancement in the presence of Al3+, which is subject to the strong quench effects caused by Cu2+ and Fe3+. The binding ratio of probe-Al3+ was determined from the Job’s plot to be 1:1. Moreover, the probe was demonstrated to be effective for in vivo imaging of the intracellular aluminum ion in both living Drosophila S2 cells and Malpighian tubules.