In computer vision, multiple object tracking (MOT) plays a crucial role in solving many important issues. A common approach of MOT is tracking by detection. Tracking by detection includes occlusions, motion prediction, and object re-identification. From the video frames, a set of detections is extracted for leading the tracking process. These detections are usually associated together for assigning the same identifications to bounding boxes holding the same target. In this article, MOT using YOLO-based detector is proposed. The authors’ method includes object detection, bounding box regression, and bounding box association. First, the YOLOv3 is exploited to be an object detector. The bounding box regression and association is then utilized to forecast the object’s position. To justify their method, two open object tracking benchmarks, 2D MOT2015 and MOT16, were used. Experimental results demonstrate that our method is comparable to several state-of-the-art tracking methods, especially in the impressive results of MOT accuracy and correctly identified detections.

In order to improve the patrol efficiency of border patrol drones, based on unmanned aerial vehicle (UAV) border patrol missions in multiple complex environments, this article proposes a whale algorithm based on chaos theory to plan patrol missions for multiple drones. First, according to the terrain the corresponding environmental model is established for the topography and then solved in layers to obtain the number of drones and other information that each base needs to send to the patrol area. Further, the use of drones with cameras and other detection equipment to patrol the scene information and images extract and transfer to the terminal in real time, and further detect suspicious persons and vehicles on the screen. The final simulation results show that the proposed scheme can be effectively applied to the planning of multi-UAV coordinated missions for border patrol.

Taiwan fish markets sell a wide variety of fish, and laypeople may have difficulty recognizing the fish species. The identification of fish species is still mostly based on illustrated handbooks, which is time-consuming when users lack experience. Automatic segmentation and recognition of fish images are important for the field of oceanography. However, in fish markets, the instability of light sources and changes in illumination influence the brightness and colors of fish. Moreover, fish markets often arrange fish together and cover them with ice to keep them fresh, thus increasing the difficulty of automatic fish recognition. This study presents a fish recognition system that combines a state-of-art instance segmentation method along with ResNet-based classification. An input image is first passed through the fish segmentation model, which crops the image into several images containing specific objects with a plain black background. Then the cropped images are assigned to a class by the fish classification model, which returns the predicted label of each image. A database of real fish images was collected from a fish market to verify the system. The experimental results revealed that the system achieved 85% Top-1 accuracy and 95% Top-5 accuracy on the test data set.

Benchmark datasets used for testing computer vision (CV) methods often contain little variation in illumination. The methods that perform well on these datasets have been observed to fail under challenging illumination conditions encountered in the real world, in particular, when the dynamic range of a scene is high. The authors present a new dataset for evaluating CV methods in challenging illumination conditions such as low light, high dynamic range, and glare. The main feature of the dataset is that each scene has been captured in all the adversarial illuminations. Moreover, each scene includes an additional reference condition with uniform illumination, which can be used to automatically generate labels for the tested CV methods. We demonstrate the usefulness of the dataset in a preliminary study by evaluating the performance of popular face detection, optical flow, and object detection methods under adversarial illumination conditions. We further assess whether the performance of these applications can be improved if a different transfer function is used.

On-demand electrohydrodynamic jetting also called electrohydrodynamic atomization (EHDA) is a method to jet small amounts of fluid out of a nozzle with a relatively large diameter by switching on and off an electrical field between the nozzle and the substrate. The total amount of volume deposited is up to 5 pL. The set-up consists of a vertically placed glass pipette with a small nozzle directed downward and a flat substrate placed close to the end of the nozzle. Inside the pipette, an electrode is mounted close to the entrance of the nozzle. The electrode is connected to a high voltage power amplifier. Upon switching on the electrical field, the apparent surface tension drops, the meniscus deforms into a cone and fluid starts to flow toward the nozzle deforming the meniscus. At a certain moment the cone reaches the Taylor cone dimensions and from its tip a jet emerges that decomposes into a stream of charged fL droplets that fly toward the substrate. This process stops when the pulse is switched off. After switching off, the meniscus returns slowly to its equilibrium position. The process is controlled by different time constants, such as the slew rate of the power amplifier and the RC time of the electrical circuit composed of the electrical resistance in the fluid contained in the nozzle between the electrode and the meniscus, and the capacitance of the gap between the meniscus and the flat substrate. Another time constant deals with the fluid flow during the growth of the meniscus, directly after switching on the pulse. This fluid flow is driven by hydrostatic pressure and opposed by a viscous drag in the nozzle. The final fluid flow during droplet formation is governed by the balance between the drag of the charge carriers inside the fluid, caused by the current associated with the charged droplets leaving the meniscus and the viscous drag. These different phenomena will be discussed theoretically and compared to experimental results.

FDM 3D printers allow massive creativity in personal products, but their potential has been limited due to inability to manipulating material properties. Previous work had demonstrated that the desired roughness could be presented simply by controlling the spatial density of tiny pins on a printed surface. This article offers a means of providing the desired softness perception of a printed surface and the desired roughness to expand the haptic dimension over which a user can exert control. Specifically, we control the softness by manipulating the infill structures of a printed surface. However, it is known that a skin contact area affects softness perception. The roughness, which is controlled by pins’ density, may also affect the perceived softness of a printed surface. Therefore, we investigate how the internal structures and the density of the pins affect softness perception. Through psychophysical experiments, we derive a computational model that estimates the perceived softness from the density of the pins and the infill density of a printed surface.

This research explored the potential for ink-jet printing to replicate the coloration and finishing techniques of traditional denim fabric and standardized the reproduction and evaluation procedure. Although denim fabric is widely consumed and very popular, one drawback to denim is that the finishing and manufacturing processes are energy and water intensive and can cause environmental hazards as well as generation of pollution through water waste, particularly at the finishing stage. Textile ink-jet printing has the potential to replicate some of the coloration and finishing techniques of traditional denim fabric without negative environmental impacts. A two-phase research project was conducted. In Phase I (P1), an optimal standard production workflow for digital denim reproduction (including color and finishing effects) was established, and six different denim samples were reproduced based on the workflow. In Phase II, an expert visual assessment protocol was developed to evaluate the acceptance of the replicated digital denim. Twelve ink-jet printing, color science, and denim industry experts finished the assessment.

Traditional depth from focus (DFF) methods obtain depth image from a set of differently focused color images. They detect in-focus region at each image by measuring the sharpness of observed color textures. However, estimating sharpness of arbitrary color texture is not a trivial task especially when there are limited color or intensity variations in an image. Recent deep learning based DFF approaches have shown that the collective estimation of sharpness in a set of focus images based on large body of training samples outperforms traditional DFF with challenging target objects with textureless or glaring surfaces. In this article, we propose a deep spatial–focal convolutional neural network that encodes the correlations between consecutive focused images that are fed to the network in order. In this way, our neural network understands the pattern of blur changes of each image pixel from a volumetric input of spatial–focal three-dimensional space. Extensive quantitative and qualitative evaluations on existing three public data sets show that our proposed method outperforms prior methods in depth estimation.

In a network information society, there are many occasions where people’s behaviors need to be tracked, photographed, and recognized. Biometric recognition technologies are considered to be one of the most effective solutions. Traditional methods mostly use graph structure and deformed component model to design two-dimensional (2D) human body component detectors, and apply graph models to establish the connectivity of each component. The recognition design process is simple, but the accuracy of recognition and tracking effect applied in monitoring image acquisition is not high. The improved particle swarm optimization algorithm is used to determine the particle structure, and the binary bit string is used to represent the particle structure. The support vector machine (SVM) parameters of discrete particles are optimized, and the synchronous optimization design of feature selection and SVM parameters is carried out to realize the synchronous optimization of portrait feature subset and SVM parameters in discrete space. Through in-depth research, the extracted feature subsets can be effectively optimized and selected, and the parameters of SVM model can be optimized synchronously. The discrete particle structure is associated with the SVM parameters to achieve feature selection and SVM parameter synchronization and optimization. It is not only superior to traditional algorithms in terms of recognition rate, but also reduces the feature dimension and shortens the recognition time. The deep feature recognition built on the learning machine is not easy to diverge and can effectively adjust the particle speed to the global optimal, which is more effective than the particle swarm algorithm to search for the global optimal solution, and has better robustness. In the experiments, the research content of the article is compared with the traditional methods to test and analysis. The results show that the method optimizes the selection of feature subset and eliminates a large number of invalid features. The method not only reduces space complexity and shortens recognition time, but also improves recognition rate. The dimension of feature subset dimensions are superior to those extracted by other algorithms.

Tone mapping is extensively researched to address the issue of displaying high dynamic range (DR) scenes on low DR displays. Even though several tone-mapping operators (TMOs) exist, not all are designed for hard real time. The operator has to be capable of scaling up the spatial resolution without compromising the frame rate. The implementation of a TMO should also be simple enough to embed in low-cost platforms for imaging systems. A computationally efficient, and well accepted, class of TMOs are global ones based on histograms. This work presents a method to implement TMOs that use histograms. This method is suitable for low-cost field-programmable gate arrays (FPGAs), using simple components such as adders, multipliers, and random access memories, and is particularly suited for a nonlinear CMOS image sensor (CIS) operating continuously in hard real time. The authors develop a fixed-point design, validated in bit-true fashion using Xilinx and Altera tools, from a background algorithm implemented using Matlab. Our generic design uses pipelined circuits and operates with low latency. The use of a hardware description language to model our circuits guarantees portability and modularity. Moreover, the complete TMO is generated from design parameters and a design template. The architecture is robust and scales well from kilopixel to megapixel formats. The circuits achieve 30 frames per second, at high definition resolutions, while occupying only a small fraction of the available logic elements in a low-cost FPGA device.