Search

This study presents dedicated hardware for iris localization that can be used as a coprocessor in the development of real-time and low-cost embedded iris biometric systems. Though the hardware architecture is described for iris localization in the visible wavelength (VW) images, the concept used can be applied to near infrared (NIR) images as well. In general, the architecture can be used for a class of iris localization algorithms based on the edge-map generation and circular Hough transform (CHT). The architecture presented here generates the edge-maps for limbic and pupil boundary detection using median filtering followed by Sobel edge detection; however, an additional reflection removal module is used for pupil boundary detection. Further, the CHT hardware module detects circle in each edge-map. The proposed architecture was implemented in programmable logic of the Zynq-7000 SoC device from Xilinx. This hardware implementation gives an iris localization accuracy of 98.43% and average processing time of 5.148 ms for UBIRIS.v1 VW database images (200 × 150 pixel). The algorithm used is suitable for less unconstrained and frontal-view iris images captured with subjects’ active participation; however, the images may contain non-ideal issues such as reflection and occlusion by eyelids and eyelashes.

In this work, a lane departure warning system (LDWS) algorithm for embedded platforms which has restricted resources is proposed. An LDWS consists of two main sub-functions which are lane detection and lane tracking. Although sophisticated methods have been developed for both sub-functions, they usually require high processing power and even GPU processing power. Therefore, they are not applicable for hardware with limited resources. In this work, Hough Transform (HT)-based lane detection algorithm is applied. The vulnerability of HT-based methods against misleading images is eliminated by the proposed filtering algorithm. Main differences of the proposed filtering algorithm from the classical methods in the literature are that it is applied in the parameter space rather than the image, and it is specialized only for determining lanes. In the lane tracking stage, the K-means clustering algorithm has been modified to operate online. In this way, the parameters of the detected lane can be followed adaptively during lane changing or overtaking. Real-time test results on embedded hardware demonstrated that the processing time does not exceed 41.67 ms with an accuracy of over 91.5%., BMC Automotive Industry Company; Ministry of Science, Industry and Technology under the title of Lane Detection and Tracking System Design for Commercial Vehicles [01229.STZ.2012-1], This project was financially supported by the BMC Automotive Industry Company and Ministry of Science, Industry and Technology under the title of Lane Detection and Tracking System Design for Commercial Vehicles and grant order No. 01229.STZ.2012-1.

The rate of road traffic accidents in the 38th and 39th week of 2020 has increased. Many factors influence the occurrence of accidents, one of the reasons is that the driver is sleepy. The parameters used to determine the condition of drowsiness are identifying the condition of the eyelids. With this identification, an image processing system can be utilized by applying the Haar Casecade Classifier and Circular Hough Transform methods with Find Countur using HSV parameters. The detection uses a camera with the driver's eye input, if the image detected is a drowsy eye, the system will wake up the driver by turning on a sound alarm, besides this system is equipped with an SMS gateway feature that will be sent to the driver's relatives, that the driver is sleepy. Message notification sent contains the condition and location of the rider. Based on system-wide testing, this system has a success rate of 100% against HSV1, 96% against HSV2 and HSV3, 89% against HSV4, and 88% against HSV5 on eye detection results and response time. The resulting error is because the lighting at a certain distance is not match with HSV set, the system experiences a freeze due to the buzzer sound delay that is too long and the performance of the PC is also less than optimal.

Iris recognition refers to identifying individuals based on iris patterns, which have been widely used in security systems, such as subway security and access control attendance, because everyone has a unique iris shape. In the study, we propose an OCaNet model for the iris recognition task. First, binarized threshold segmentation is used to locate the pupil and the pupil boundary is obtained; then, the Hough transform is applied to locate the outer edge of the iris; according to the located pupil and iris, the iris area image is obtained through image segmentation; finally, the iris image is normalized to adjust each original image to the same size and corresponding position, so as to eliminate the influence of translation, scaling, and rotation on iris recognition. Second, the normalized iris images are both input into the octave convolution module and attention module. The octave convolution module is used to extract the shape and contour features of the iris by decomposing the feature map into high and low frequencies. The attention module is applied to extract the color and texture characteristics of the iris. Finally, the two feature maps are concatenated and produce a distribution of output classes. Experimental results show that the proposed OCaNet model is significantly more accurate.

Image retrieval system has numerous applications in different domains and generated promising results in each domain in case of single panel images. However, in addition to single panel images, the use of multi-panel images particularly in medical and research domains is increasing tremendously. The image retrieval system treats the multi-panel image as a single image and has no potential to get access to its sub-images, as a result the retrieval accuracy of image retrieval system is affected. The latest sub-image separation methods use edge image for locating the position of lines in the input multi-panel image. The edge image of the low contrast multi-panel image often contains broken lines. These broken lines in the edge image cause misdetection of line(s) in the input multi-panel image which in turn affect the accuracy of sub-image separation. Furthermore, their results are not satisfactory for large scale multi-panel images (i.e., multi-panel images having more than seven sub-images along one of its two sides) and multi-panel images having no solid border line(s). In this paper, an effective sub-image separation method for improving the accuracy of sub-image separation is proposed. The proposed method bridges the gaps in each line of the input image using Hough transform. Next, the positions of all lines (i.e., border lines and sub-image separators) in each image are determined and the true border lines among them are identified using a simple border identification model and detached. A fast sub-image separation method is then employed for separating the sub-images of multi-panel image using the true sub-image separators. We tested our proposed method on testing dataset containing 2225 images. The experimental results show that our proposed method outperforms the latest methods.

Hand gesture is a natural interaction method, and hand gesture recognition is familiar in human–computer interaction. Yet, the variations, as well as the complexity of hand gestures such as self-structural characteristics, views, and illuminations, made hand gesture recognition as a challenging task. Nowadays, the human–computer interaction area enhancement leads to putting interest in the dynamic hand gesture segmentation based on the gesture recognition system. Apart from the lengthy clinical success, dynamic hand gesture segmentation through webcam vision seems challenging due to the light effects, partial occlusion, and complicated environment. Hence, to segment the entire hand gesture region and enhance the segmentation accuracy, this paper develops an improved segmentation and deep learning-based strategy for dynamic hand gesture recognition. The data is gathered from the ISL benchmark dataset that consists of both static as well as dynamic images. The initial process of the proposed model is the pre-processing, which is being performed by grey scale conversion and histogram equalization. Further, the segmentation of gestures is done by the novel Adaptive Hough Transform (AHT), where the theta angle is tuned. Once the segmentation of gestures is done, the optimized Deep Convolutional Neural Network (Deep CNN) is used for gesture recognition. The learning rate, epoch count, and hidden neurons are tuned by the same heuristic concept. As the main contribution, the segmentation and classification are enhanced by the hybridization of Electric Fish Optimization (EFO), and Whale Optimization Algorithm (WOA) called Electric Fish-based Whale Optimization Algorithm (E-WOA). The training of optimized Deep CNN is handled by Dynamic Time Warping (DTW) for avoiding redundant frames, thus enhancing the performance of dynamic hand gestures. Quantitative measurement is accomplished for evaluating hand gesture segmentation and recognition, which portrays the superior behaviour of the proposed model.

The perception, localization, and navigation of its environment are essential for autonomous mobile robots and vehicles. For that reason, a 2D Laser rangefinder sensor is used popularly in mobile robot applications to measure the origin of the robot to its surrounding objects. The measurement data generated by the sensor is transmitted to the controller, where the data is processed by one or multiple suitable algorithms in several steps to extract the desired information. Universal Hough Transform (UHT) is one of the appropriate and popular algorithms to extract the primitive geometry such as straight line, which later will be used in the further step of data processing. However, the UHT has high computational complexity and requires the so-called accumulator array, which is less suitable for real-time applications where a high speed and low complexity computation is highly demanded. In this study, an Accumulator-free Hough Transform (AfHT) is proposed to reduce the computational complexity and eliminate the need for the accumulator array. The proposed algorithm is validated using the measurement data from a 2D laser scanner and compared to the standard Hough Transform. As a result, the extracted value of AfHT shows a good agreement with that of UHT but with a significant reduction in the complexity of the computation and the need for computer memory.

Unmanned floating vehicles (UFVs) can play a major role in the removal of invasive floating aquatic weeds in the water bodies. To provide automation to those UFVs, image or video processing techniques are helpful. Waterline detection from the sky–land–water region is the primary step for UFV autonomous navigation. After detecting the water region, object detection in the water region is the secondary step. In this paper, an object detection algorithm in water for low cost UFVs is proposed. Real-time and accurate waterline detection is done by using the K-means algorithm and Fast Marching method and it is compared with the Hough transform. Objects are detected accurately in the water region by a modified gradient-based image processing algorithm. In this algorithm, instead of Sobel Gradient Operator, Finite Difference Central Gradient Operator is used to achieve high accuracy in detecting objects in the water region. Floating green color weeds are also detected from objects using the green color of the weeds, and weed percentage is also calculated. Experiments were done on Own datasets, and OSF datasets containing freely floating plants, small cluster plants, and birds. Raspberry Pi 3 Model-B processor is used to implement this proposed algorithm. These performance metrics results show that the proposed detection approach is effective, more accurate, and suitable to different lake environments.

Identifying and categorizing contours in images is important in many areas of computer vision. Examples include such operational tasks solved by using unmanned aerial vehicles as dynamic monitoring of the condition of transport infrastructure, in particular road markings. This study has established that current methods of image contour analysis do not produce clear and reliable results when solving the task of monitoring the state of road markings. Therefore, it is a relevant scientific and applied task to improve the methods and models of filtration, processing of binary images, and qualitative and meaningful separation of the boundaries of objects of interest. To solve the task of highlighting road marking contours on images acquired from an unmanned aerial vehicle, a method has been devised that includes an operational tool for image preprocessing – a combined filter. The method has several advantages and eliminates the limitations of known methods in determining the boundaries of the location of the object of interest, by highlighting the contours of a cluster of points using histograms. The method and procedures reported here make it possible to successfully solve problems that are largely similar to those that an expert person can face when solving intelligent tasks of processing and filtering information. The proposed method was verified at an enterprise producing the Ukrainian unmanned aerial vehicle "Spectator" during tests of information technology of dynamic monitoring of the state of transport infrastructure. The results could be implemented in promising intelligent control systems in the field of modeling human conscious behavior when sorting data required for the perception of environmental features

The dropper is one of the core components of the high speed railway catenary and dropper failure will lead to serious transportation accidents. It is very important to carry out dropper failure and defect detection. There are a large number of droppers installed in the catenary. The dropper images are collected by the high-definition camera installed on the top of the moving catenary inspection vehicles. The image quality and image consistency are poor, and it is very difficult to identify dropper defects automatically. The railway department and companies lack efficient and intelligent detection methods. This article innovatively proposes a multialgorithm fusion image processing technology, and builds a dropper recognition and failure–defect detection model based on a deep learning algorithm and subpixel level dropper defect detection model, and achieves high accuracy dropper failure and defect detection. The detection model based on the Faster R-CNN algorithm is studied to realize the positioning and recognition of the dropper and the failure detection of bending slack and broken dropper. The subpixel level dropper defect detection algorithm is based on the fusion of the image preprocessing, dropper fine positioning algorithm, edge fitting and bending zoom algorithms, the Hough transform algorithm, and so on. These can be used to realize detection of defects, such as microdeformation, dropper-strands loosened, dropper-strands broken, and foreign body adhesions. The test is verified by the catenary images taken from a practical high speed railway. The detection accuracy, real-time performance, and stability of the algorithm meet the needs of inspection and maintenance for a high speed railway.

Sports video is loved by the audience because of its unique charm, so it has high research value and application value to analyze and study the video data of competition. Based on the background of football match, this paper studies the football detection and tracking algorithm in football game video and analyzes the real-time image of real-time mobile devices in sports video augmented reality. Firstly, the image is preprocessed by image graying, image denoising, image binarization, and so on. Secondly, Hough transform is used to locate and detect football, and according to the characteristics of football, Hough transform is improved. Based on the good performance of SIFT algorithm in feature matching, a football tracking algorithm based on SIFT feature matching is proposed, which matches the detected football with the sample football. The simulation results show that the improved Hough transform can effectively detect football and has good antijamming performance. And the designed football tracking algorithm based on SIFT feature matching can accurately track the football trajectory; therefore, the football detection and tracking algorithm designed in this paper is suitable for real-time football monitoring and tracking.

The internal structure analysis of radial tires is of great significance to improve vehicle safety and during tire research. In order to perform the digital analysis and detection of the internal composition in radial tire cross-sections, a detection method based on digital image processing was proposed. The research was carried out as follows: (a) the distribution detection and parametric analysis of the bead wire, steel belt, and carcass in the tire section were performed by means of digital image processing, connected domain extraction, and Hough transform; (b) using the angle of location distribution and area relationship, the detection data were optimized through coordinate and quantity relationship constraints; (c) a detection system for tire cross-section components was designed using the MATLAB platform. Our experimental results showed that this method displayed a good detection performance, and important practical significance for the research and manufacture of tires.

Purpose The purpose of this study is to automate the slice thickness verification on the AAPM CT performance phantom and validate it for variations of slice thickness, position from iso‐center, and reconstruction filter. Methods An automatic procedure for slice thickness verification on AAPM CT performance phantom was developed using MATLAB R2015b. The stair object image within the phantom was segmented, and the middle stair object was located. Its angle was determined using the Hough transformation, and the image was rotated accordingly. The profile through this object was obtained, and its full‐width of half maximum (FWHM) was automatically measured. The FWHM indicated the slice thickness of the image. The automated procedure was applied with variations in three independent parameters, i.e., the slice thickness, the distance from the phantom to the iso‐center, and the reconstruction filter. The automated results were compared to manual measurements made using electronic calipers. Results The differences of the automated results from the nominal slice thicknesses were within 1.0 mm. The automated results are comparable to those from manual approach (i.e., the difference of both is within 12%). The automatic procedure accurately obtained slice thickness even when the phantom was moved from the iso‐center position by up to 4 cm above and 4 cm below the iso‐center. The automated results were similar (to within 0.1 mm) for various reconstruction filters. Conclusions We successfully developed an automated procedure of slice thickness verification and confirmed that the automated procedure provided accurate results. It provided an easy and effective method of determining slice thickness.

Perception of the environment is the prerequisite for the realization of unmanned driving. Correctly detecting the lane line and navigating the vehicle is the key technology in unmanned driving. This paper mainly aims at the low accuracy of the traditional lane detection algorithm in the complex environment such as night and rain, and proposes a lane detection and recognition method based on dynamic region of interest (ROI) selection and firefly algorithm. First, perform distortion correction on the captured lane image, gray scale and blur image preprocessing, and then determine the height of the ROI based on the vanishing point, and dynamically adjust the width of the ROI based on the recognition of the lane line in the previous frame to achieve dynamic ROI adjust to eliminate interference factors and reduce the amount of calculation to the greatest extent. Finally, to solve the problem that the canny operator is sensitive to noise in the traditional method, an improved firefly algorithm is proposed for edge detection. The slope-limited progressive probability Hough transform is used to detect the straight line of the ROI divided into several boxes, and the least square method is used to fit several detected straight lines to extract lane lines. Experimental results show that the method we proposed can achieve lane line detection well in complex environments, with an average accuracy rate of 96.37%, and an average detection time per frame of only 118 ms.

Pupil detection in a human eyeimage or video plays a key role in many applications such as eye-tracking, diabetic retinopathy screening, smart homes, iris recognition, etc. Literature reveals pupil detection faces many complications including light reflections, cataract disease, pupil constriction/dilation moments, contact lenses, eyebrows, eyelashes, hair strips, and closed eye. To cope with these challenges, research community has been struggling to devise resilient pupil localization schemes for the image/video data collected using the near-infrared (NIR) or visible spectrum (VS) illumination. This study presents a critical review of numerous pupil detection schemes taken from standard sources. This review includes pupil localization schemes based on machine learning, histogram/thresholding, Integro-differential operator (IDO), Hough transform and among others. The probable pros and cons of each scheme are highlighted. Finally, this study offers recommendations for designing a robust pupil detection system. As scope of pupil detection is very broader, therefore this review would be a great source of information for the relevant research community.

Modern unmanned aerial vehicles (UAVs), including combat drones, are usually equipped with cameras for imaging. A camera is mainly used to increase the accuracy of the attack by ensuring stable-guided flight toward the target when the UAV detects the target in the terminal phase. Despite the high cost of these pieces of equipment, their utilization is in fact low, except during the final stage. The method proposed here extends the range of use of a UAV-mounted camera to the navigation stage. When GPS is unable to estimate the position due to jamming or a sensor error, the possibility of using the position estimation value calculated through vision navigation is described.

This article presents a novel target detection method, which is named as circular target feature detection framework based on a deep convolutional neural network (DCNN). The central proposition of this method uses the optimized DCNN architecture to detect the target and locate the position of the circle accurately in the image field of view. In this article, a Hough transform based on threshold processing (HTP) is embedded into the optimized DCNN architecture, which calculates the center positions and radius of all circles by training the circular samples for each detected rectangular frame. It can efficiently identify small circular target materials in the industry and screen out unqualified particles. The experimental results show that the boundary information of the circles is obtained clearly from the complex noise background images, thereby accurately determining the location of the circle. It has some advantages over only using a specific circular recognition algorithm. We proposed the new study on HTP-DCNN, which has extremely high accuracy in the field of machine vision positioning with circles for industrial applications.

Extraction of news text captions aims at a digital understanding of what is happening in a specific region during a certain period that helps in better communication between different nations because we can easily translate the plain text from one language to another. Moving text captions causes blurry effects that are a significant cause of text quality impairments in the news channels. Most of the existing text caption detection models do not address this problem in a way that captures the different dynamic motion of captions, gathers a full news story among several frames in the sequence, resolves the blurring effect of text motion, offers a language-independent model, or provides it as an end-to-end solution for the community to use. We process the frames coming in sequence and extract edge features using either the Hough transform or our color-based technique. We verify text existence using a Convolutional Neural Network (CNN) text detection pre-trained model. We analyze the caption motion status using hybrid pre-trained Recurrent Neural Network (RNN) of Long Short-Term Memory (LSTM) type model and correlation-based model. In case the motion is determined to be horizontal rotation, there are two problems. First, it means that text keeps rotating with no stop resulting in a high blurring effect that affects the text quality and consequently resulting in low character recognition accuracy. Second, there are successive news stories which are separated by the channel logo or long spaces. We managed to solve the first problem by deblurring the text image using either Bicubic Spline Interpolation (BSI) technique or the Denoising Autoencoder Neural Network (DANN). We solved the second problem using a Point Feature Matching (PFM) technique to match the existing channel logo with the channels’ logo database (ground truth). We evaluate our framework using Abbyy® SDK as a standalone tool used for text recognition supporting different languages.

The visual inspection algorithms are the core of automatic optical inspection system on surface mounting machines. This article is concerned with the development of precise positioning algorithms for circular mark points and transistor (TR) components on surface mounting devices. To handle nonuniform illumination or occlusion of other components, a polar coordinate transform and smoothness selection based circular mark point location method is proposed. The TR components are fundamental chips in electronic products and have various package types. The illumination changes, background disturbance, and the diversity of package types have imposed great challenges on the development of the uniform algorithm for detection and location of TR components. To deal with these issues, the 1-D integral image based TR component detection and location algorithm is proposed and the coordinates and orientation of the component are calculated simultaneously. The efficiency of the proposed methods is tested on real images and compared with classical Hough transform method, commercial algorithms on SMT482 device, and two methods of Halcon software.

A new fast method for pupil detection and eyetracking real time is being developed based on the study of a boundary-step model of a grayscale image by the Laplacian-Gaussian operator and finding a new proposed descriptor of accumulated differences (point identifier), which displays a measure of the equidistance of each point from the boundaries of some relative monotonous area (for example, the pupil of the eye). The operation of this descriptor is based on the assumption that the pupil in the frame is the most rounded monotonic region with a high brightness difference at the border, the pixels of the region should have an intensity less than a predetermined threshold (but the pupil may not be the darkest region in the image). Taking into account all of the above characteristics of the pupil, the descriptor allows achieving high detection accuracy of its center and size, in contrast to methods based on threshold image segmentation, based on the assumption of the pupil as the darkest area, morphological methods (recursive morphological erosion), correlation or methods that investigate only the boundary image model (Hough transform and its variations with two-dimensional and three-dimensional parameter spaces, the Starburst algorithm, Swirski, RANSAC, ElSe). The possibility of representing the pupil tracking problem as a multidimensional unconstrained optimization problem and its solution by the Hook-Jeeves non-gradient method, where the function expressing the descriptor is used as the objective function, is investigated. In this case, there is no need to calculate the descriptor for each point of the image (compiling a special accumulator function), which significantly speeds up the work of the method. The proposed descriptor and method were analyzed, and a software package was developed in Python 3 (visualization) and C ++ (tracking kernel) in the laboratory of the Physics and Mathematics Faculty of Kamchatka State University of Vitus Bering, which allows illustrating the work of the method and tracking the pupil in real time.

Tilt correction is an essential step in the license plate recognition system (LPR). The main goal of this article is to provide a review of the various methods that are presented in the literature and used to correct different types of tilt that appear in the digital image of the license plates (LP). This theoretical survey will enable the researchers to have an overview of the available implemented tilt detection and correction algorithms. That’s how this review will simplify for the researchers the choice to determine which of the available rotation correction and detection algorithms to implement while designing their LPR system. This review also simplifies the decision for the researchers to choose whether to combine two or more of the existing algorithms or simply create a new efficient one. This review doesn’t recite the described models in the literature in a hard-narrative tale, but instead, it clarifies how the tilt correction stage is divided based on its initial steps. The steps include: locating the plate corners, finding the tilting angle of the plate, then, correcting its horizontal, vertical, and sheared inclination. For the tilt correction stage, this review clarifies how state-of-the-art literature handled each step individually. As a result, it has been noticed that line fitting, Hough transform, and Randon transform are the most used methods to correct the tilt of a LP.

A localization system using reflective markers and a fisheye camera with blinking infrared lights is useful and safe for mobile robot navigation in an environment with coexisting humans and robots; however, it has the problems of low robustness and a small measurable range for marker detection. A large, square-shaped reflective marker, with solid and dotted edges, is proposed for more reliable localization of indoor mobile robots. It can be easily detected using Hough transform and is robust for occlusion. The coordinates of the four corners of the square-shaped marker determine the robot’s localization. Infrared lighting with a new LED arrangement is designed for a wide measurable range via brightness simulation, including the effect of observation and reflection angles. A prototype system was developed, enabling the 2D position and orientation to be detected with an accuracy of 60 mm and 3◦, respectively, within a 4 m2 area.

This paper addresses the challenge of reading low contrast text on tyre sidewall images of vehicles in motion. It presents first of its kind, a full scale industrial system which can read tyre codes when installed along driveways such as at gas stations or parking lots with vehicles driving under 10 mph. Tyre circularity is first detected using a circular Hough transform with dynamic radius detection. The detected tyre arches are then unwarped into rectangular patches. A cascade of convolutional neural network (CNN) classifiers is then applied for text recognition. Firstly, a novel proposal generator for the code localization is introduced by integrating convolutional layers producing HOG-like (Histogram of Oriented Gradients) features into a CNN. The proposals are then filtered using a deep network. After the code is localized, character detection and recognition are carried out using two separate deep CNNs. The results (accuracy, repeatability and efficiency) are impressive and show promise for the intended application.

Aiming at the issue of incomplete trajectories in the 2D epipolar image of circular light field, this paper proposes a 3D reconstruction method by using 3D Hough transformation. This method computes 3D point clouds by computing the parameters of feature trajectories in 3D image volume. By analyzing the 3D distribution of circular light field trajectories, binary curves in image volume are extracted, and their local orientation are further estimated by the 3D structure tensor. The 3D Hough space generation and the parameter selection method are proposed to the 3D curves detection. The parameters of these curves are converted to 3D point clouds on each view and then merged to final 3D reconstruction. The ambiguity of Hough transformation solution on 2D epipolar image is overcome by the 3D analyzing method. The experiments are carried out on both synthetic and real datasets. The experiment results show that this method can improve the reconstruction performance compared with the state-of-the-art in circular light field.

This paper proposes a design of a state estimator for tracking and predicting the object trajectory for the manipulation using a dexterous multi-fingered Delta robot. The observations of the object state acquired from the cameras (Basler), in the real-time scenario. Initially, pixels are removed that corresponds to the background pixels using a mixture of Gaussian algorithms. Secondly, the color invariant approach is implemented as a Hough transform. The same is used for the tracking of the object. This results in the color invariant thresholding to filter the region of interest. As the successive frames have some noise, morphological operations have also performed in to remove if any present outlier. After removing the noise from the frame, estimating the object center followed by velocity estimation done using the k-means clustering. Kalman predictor is used for the prediction of the future state(s) using the current state and known system dynamics. The catching strategy of the object using the Delta robot-based multi-fingered architecture is also discussed. Different trajectories and objects are provided for the catching of the object.

In the last decades, the automation process has improved humans’ daily life. As technology continues to evolve, software and machines are increasingly being used to automate everyday tasks that were once performed by humans. For instance, sports have already seen some use of wearable devices in activities involving monitoring and measurement of position speed and body parameters, for example, heart rate and maximal oxygen consumption (VO2). Despite that, a personal trainer is still necessary to catch athlete’s errors. Computer vision can be applied to this problem, helping athletes to improve their performances. Therefore, this work describes a low-cost computer vision algorithm to track a tennis ball’s vertical motion. The algorithm aims to determine the touch position and to record a picture of the moment for analysis. The method works by applying color filters and transformations combined with a support vector machine to accurately detect the line and ball positions. The ball position is tracked along with different pictures, and a mathematical model is used to estimate the ball trajectory and helping predict the touching instant. An Extended Kalman Filter is applied in this work to reduce measurement error and improve detection performance by predicting future positions for the ball. The ground line positions are determined based on a Hough transform to detect the tennis court lines. The results have shown excellent performance and technical feasibility of the method for deployment in low cost embedded computers.