Saliency Detection with Moving Camera via Background Model Completion

Detecting saliency in videos is a fundamental step in many computer vision systems. Saliency is the significant target(s) in the video. The object of interest is further analyzed for high-level applications. The segregation of saliency and the background can be made if they exhibit different visual cues. Therefore, saliency detection is often formulated as background subtraction. However, saliency detection is challenging. For instance, dynamic background can result in false positive errors. In another scenario, camouflage will result in false negative errors. With moving cameras, the captured scenes are even more complicated to handle. We propose a new framework, called saliency detection via background model completion (SD-BMC), that comprises a background modeler and a deep learning background/foreground segmentation network. The background modeler generates an initial clean background image from a short image sequence. Based on the idea of video completion, a good background frame can be synthesized with the co-existence of changing background and moving objects. We adopt the background/foreground segmenter, which was pre-trained with a specific video dataset. It can also detect saliency in unseen videos. The background modeler can adjust the background image dynamically when the background/foreground segmenter output deteriorates during processing a long video. To the best of our knowledge, our framework is the first one to adopt video completion for background modeling and saliency detection in videos captured by moving cameras. The F-measure results, obtained from the pan-tilt-zoom (PTZ) videos, show that our proposed framework outperforms some deep learning-based background subtraction models by 11% or more. With more challenging videos, our framework also outperforms many high-ranking background subtraction methods by more than 3%.

Research on Population Development Trend in Huizhou of China Forecast Based on Optimal Weighted Combination Method and Fractional Grey Model

In this paper, the optimal weighted combination model and fractional grey model are constructed. The coefficients of the optimal weighted combination model are determined by minimizing the sum of squares of resists of each model. On the other hand, the optimal conformable fractional order and dynamic background value coefficient are determined by the quantum inspired evolutionary algorithm (QIEA). Taking the resident population from 2008 to 2018 as the research object, the optimal weighted combination model and fractional grey model were used to study the estimated and predicted values. The results are compared and analyzed. The results show that the fractional grey model is better than the optimal weighted combination model in the estimation of the values. The optimal weighted combination model is better than the fractional grey model in predicting. Meanwhile, the fractional grey model is found to be very suitable for the data values that are large, and the changes between the data are relatively small. The research results expand the application of the fractional grey model and have important implications for the policy implementation activities of Huizhou government according to the population growth trend in Huizhou.

Luminescence of Baikal water as a dynamic background of the Baikal-GVD Neutrino Telescope

The Baikal-GVD is a neutrino telescope situated in the deepest freshwater lake in the world — Lake Baikal. The design of the Baikal-GVD trigger system allows also to study the ambient light of the lake. The analysis of the optical light activity of Baikal water, particularly, time and spatial variations of the luminescence activity for data collected in years 2018, 2019, and 2020 is presented. For the first time we observed highly luminescent layer moving upwards with maximal speed of 28 m/day in January 2021.

Detection of the Deep-Sea Plankton Community in Marine Ecosystem with Underwater Robotic Platform

Variations in the quantity of plankton impact the entire marine ecosystem. It is of great significance to accurately assess the dynamic evolution of the plankton for monitoring the marine environment and global climate change. In this paper, a novel method is introduced for deep-sea plankton community detection in marine ecosystem using an underwater robotic platform. The videos were sampled at a distance of 1.5 m from the ocean floor, with a focal length of 1.5–2.5 m. The optical flow field is used to detect plankton community. We showed that for each of the moving plankton that do not overlap in space in two consecutive video frames, the time gradient of the spatial position of the plankton are opposite to each other in two consecutive optical flow fields. Further, the lateral and vertical gradients have the same value and orientation in two consecutive optical flow fields. Accordingly, moving plankton can be accurately detected under the complex dynamic background in the deep-sea environment. Experimental comparison with manual ground-truth fully validated the efficacy of the proposed methodology, which outperforms six state-of-the-art approaches.

Moving-object detection and tracking by scanning LiDAR mounted on motorcycle based on dynamic background subtraction

This paper presents a method for moving-object detection and tracking (DATMO) in global navigation satellite systems (GNSS)-denied environments using a light detection and ranging (LiDAR) mounted on a motorcycle. Distortion in the scanning LiDAR data is corrected by estimating the pose (3D positions and attitude angles) of the motorcycle in a period shorter than the LiDAR scan period using normal distributions transform-based simultaneous localization and mapping (NDT-based SLAM) and the information from an inertial measurement unit (IMU) via the extended Kalman filter (EKF). The scan data of interest are extracted by subtracting the local environment map generated by NDT-based SLAM from the LiDAR scan data. Moving objects are detected from the scan data of interest using an occupancy grid method and are tracked with a Bayesian filter. Experimental results obtained from public road and university campus environments demonstrate the effectiveness of the proposed method.

Motion: enhancing signals and concealing cues

ABSTRACT Animal colour patterns remain a lively focus of evolutionary and behavioural ecology, despite the considerable conceptual and technical developments over the last four decades. Nevertheless, our current understanding of the function and efficacy of animal colour patterns remains largely shaped by a focus on stationary animals, typically in a static background. Yet, this rarely reflects the natural world: most animals are mobile in their search for food and mates, and their surrounding environment is usually dynamic. Thus, visual signalling involves not only animal colour patterns, but also the patterns of animal motion and behaviour, often in the context of a potentially dynamic background. While motion can reveal information about the signaller by attracting attention or revealing signaller attributes, motion can also be a means of concealing cues, by reducing the likelihood of detection (motion camouflage, motion masquerade and flicker-fusion effect) or the likelihood of capture following detection (motion dazzle and confusion effect). The interaction between the colour patterns of the animal and its local environment is further affected by the behaviour of the individual. Our review details how motion is intricately linked to signalling and suggests some avenues for future research. This Review has an associated Future Leader to Watch interview with the first author.

Privacy-Oriented Successive Approximation Image Position Follower Processing

In this paper, we analyze the location-following processing of the image by successive approximation with the need for directed privacy. To solve the detection problem of moving the human body in the dynamic background, the motion target detection module integrates the two ideas of feature information detection and human body model segmentation detection and combines the deep learning framework to complete the detection of the human body by detecting the feature points of key parts of the human body. The detection of human key points depends on the human pose estimation algorithm, so the research in this paper is based on the bottom-up model in the multiperson pose estimation method; firstly, all the human key points in the image are detected by feature extraction through the convolutional neural network, and then the accurate labelling of human key points is achieved by using the heat map and offset fusion optimization method in the feature point confidence map prediction, and finally, the human body detection results are obtained. In the study of the correlation algorithm, this paper combines the HOG feature extraction of the KCF algorithm and the scale filter of the DSST algorithm to form a fusion correlation filter based on the principle study of the MOSSE correlation filter. The algorithm solves the problems of lack of scale estimation of KCF algorithm and low real-time rate of DSST algorithm and improves the tracking accuracy while ensuring the real-time performance of the algorithm.