Wave Measurements Using Multi-Frame Processing of Marine Radar Data

Marine radars have proven to be useful for measuring ocean waves, but the accuracy of the measurements is limited by several factors including the look-angle dependence of the radar signals as well as noise in the radar data. The look-angle dependence introduces a systematic error or bias in the measurements, and noise causes a random error. This paper describes a method of combining data from multiple radar frames that is optimal in the sense of minimizing the error for a set of biased measurements with random additive noise. The results are shown experimentally to increase the correlation of the radar estimates with buoy measurements.

DEEP LEARNING TECHNOLOGY FOR VIDEOFRAME PROCESSING IN FACE SEGMENTATION ON MOBILE DEVICES

The aim of the research is to reduce the frame processing time for face segmentation on videos on mobile devices using deep learning technologies. The paper analyzes the advantages and disadvantages of existing segmentation methods, as well as their applicability to various tasks. The existing real-time realizations of face segmentation in the most popular mobile applications, which provide the functionality for adding visual effects to videos, were compared. As a result, it was determined that the classical segmentation methods do not have a suitable combination of accuracy and speed, and require manual tuning for a particular task, while the neural network-based segmentation methods determine the deep features automatically and have high accuracy with an acceptable speed. The method based on convolutional neural networks is chosen for use because, in addition to the advantages of other methods based on neural networks, it does not require such a significant amount of computing resources during its execution. A review of existing convolutional neural networks for segmentation was held, based on which the DeepLabV3+ network was chosen as having sufficiently high accuracy and being optimized for work on mobile devices. Modifications were made to the structure of the selected network to match the task of two classes segmentation and to speed up the work on devices with low performance. 8-bit quantization was applied to the values processed by the network for further acceleration. The network was adapted to the task of face segmentation by transfer learning performed on a set of face images from the COCO dataset. Based on the modified and additionally trained segmentation model, a mobile app was created to record video with real-time visual effects, which applies segmentation to separately add effects on two zones - the face (color filters, brightness adjustment, animated effects) and the background (blurring, hiding, replacement with another image). The time of frames processing in the application was tested on mobile devices with different technical characteristics. We analyzed the differences in testing results for segmentation using the obtained model and segmentation using the normalized cuts method. The comparison reveals a decrease of frame processing time on the majority of devices with a slight decrease of segmentation accuracy.

Just “performance nonsense”?: How recipients process news photos of activists’ symbolic actions about climate change politics

In this article, I investigate how recipients make sense of images that show symbolic actions by environmental activists during two recent United Nations Climate Change Conferences. Environmental advocacy groups are successful in creating visibility for their symbolic actions via news visuals, but little empirical evidence exists about how ordinary media recipients engage with this type of imagery. Can they understand the intended meaning of complex visual rhetoric used by environmental activists? I use think-aloud protocols to uncover the cognitive strategies which are used in processing these stylised visual claims. Results show that news photos rarely manage to communicate the intended meaning of symbolic actions. By systematically analysing various stages of visual frame processing, this study offers insights into specific configurations of the image-viewer relationship that cause high levels of ambiguity and prevent staged visual claims from being understood as intended. Yet I also find empirical evidence for a visual framing approach that works well and describe this recipe for effective communication via symbolic action photography.

Smart and real time image dehazing on mobile devices

Haze is one of the common factors that degrades the visual quality of the images and videos. This diminishes contrast and reduces visual efficiency. The ALS (Atmospheric light scattering) model which has two unknowns to be estimated from the scene: atmospheric light and transmission map, is commonly used for dehazing. The process of modelling the atmospheric light scattering is complex and estimation of scattering is time consuming. This condition makes dehazing in real-time difficult. In this work, a new approach is employed for dehazing in real-time which reads the orientation sensor of mobile device and compares the amount of rotation with a pre-specified threshold. The system decides whether to recalculate the atmospheric light or not. When the amount of rotation is little means there are only subtle changes to the scene, it uses the pre-estimated atmospheric light. Therefore, the system does not need to recalculate it at each time instant and this approach accelerates the overall dehazing process. 0.07s fps (frame per second) per frame processing time (~15fps) is handled for 360p imagery. Frame processing time results show that our approach is superior to the state-of-the-art real-time dehazing implementations on mobile operating systems.

Ultrasonic software-hardware complex for tracking public transport passengers

The article provides the description of an on-board no-touch software-hardware complex for tracking public transport passengers. The complex was designed by the authors. It allows carrying out the following actions in real time: writing down the readings of an ultrasound range finder on objects entering and exiting the door frame; processing the accumulated sets of data and determining the number of people that entered and exited the vehicle during stops; sending out the data through standard connection modules of the vehicle in order to provide specialized monitoring services with information on passenger flow, as well as solving corresponding tasks of predictive analytics. The article overviews existing analogs that are based on using laser detectors, thermal scopes, computer vision technology and floor impact detectors. Their advantages as well as weaknesses that limit their mass use are highlighted. The article also provides the following for the offered software-hardware complex: comparison results, grounds for implementing it on public transport; a description of the hardware part – the composition of the main accessories used (an STM32 controller, an ultrasound diastimeter, an infrared diastimeter); methods for processing the data given by the ultrasound range finder in order to determine the number of passengers that entered and exited the vehicle, including situations when there are many of them on both sides of the door frame, including alternating motions to enter and exit in groups as well as one by one; options of module performance according to the case’s form factor, as well as according to compositions and the number of accessories used; options of compiling modules and describing their installation schemes according to the specifics of the tasks of use to control door frames of various width, including random width.

Real-time Image Contrast Enhancement VLSI Design for Intelligent Autonomous Vehicles

A novel hardware-oriented image contrast enhancement algorithm is proposed in this study for intelligent autonomous vehicles. It utilizes a weighted filter and calculates the brightness values of an image based on the adjusted image. The brightness values are processed to either reduce or increase the brightness values of the points. To further improve the quality of an image, the algorithm implements a block-based pixel processing as opposed to a per image frame processing. The brightness values for each block or area in the image are used to improve the contrast of the image. This is accomplished by reducing or increasing the different brightness values of the pixel or lifting point in each block. Simulation results showed that compared with previously proposed algorithms, this work improved on the average discrete entropy by 1% and increased the average color enhancement factor by 8.5%. The proposed novel algorithm was realized using TSMC 0.18 μm CMOS cell process. The VLSI design has a total gate count of 6028 and operates with a frequency of 201 MHz and a power rating of 17.47 mW.

Interaction of Liquid Droplets in Gas and Vapor Flows

We investigated the conditions, characteristics, and outcomes of liquid droplet interaction in the gas medium using video frame processing. The frequency of different droplet collision outcomes and their characteristics were determined. Four interaction regimes were identified: bounce, separation, coalescence, and disruption. Collision regime maps were drawn up using the Weber, Reynolds, Ohnesorge, Laplace, and capillary numbers, as well as dimensionless linear and angular parameters of interaction. Significant differences were established between interaction maps under ideal conditions (two droplets colliding without a possible impact of the neighboring ones) and collision of droplets as aerosol elements. It was shown that the Weber number could not be the only criterion for changing the collision mode, and sizes and concentration of droplets in aerosols influence collision modes. It was established that collisions of droplets in a gaseous medium could lead to an increase in the liquid surface area by 1.5–5 times. Such a large-scale change in the surface area of the liquid significantly intensifies heat transfer and phase transformations in energy systems.