Search results

The purpose of this paper is to develop a method for a vision-based automatic landing of a multi-rotor unmanned aerial vehicle (UAV) on a moving platform. The landing system must be highly accurate and meet the size, weigh, and power restrictions of a small UAV.

The vision-based landing system consists of a pattern of red markers placed on a moving target, an image processing algorithm for pattern detection, and a servo-control for tracking. The suggested approach uses a color-based object detection and image-based visual servoing.

The developed prototype system has demonstrated the capability of landing within 25 cm of the desired point of touchdown. This auto-landing system is small (100×100 mm), light-weight (100 g), and consumes little power (under 2 W).

The novelty and the main contribution of the suggested approach are a creative combination of work in two fields: image processing and controls as applied to the UAV landing. The developed image processing algorithm has low complexity as compared to other known methods, which allows its implementation on general-purpose low-cost hardware. The theoretical design has been verified systematically via simulations and then outdoors field tests.

The purpose of this paper is to explore the challenges faced by the automatic recognition systems over the conventional systems by implementing a novel approach for detecting and recognizing the vehicle license plates in order to increase the security of the vehicles. This will also increase the societal discipline among vehicle users.

From a methodological point of view, the proposed system works in three phases which includes the pre-processing of the input image from the database, applying segmentation to the processed image, and finally extracting and recognizing the image of the license plate.

The proposed paper provides an analysis that demonstrates the correctness of the algorithm to correctly capture the license plate using performance metrics such as detection rate and false positive rate. The obtained results demonstrate that the proposed algorithm detects vehicle license plates and provides detection rate of 93.34 percent with false positive rate of 6.65 percent.

The proposed license plate detection system eliminates the need of manually used systems for managing the traffic by installing the toll-booths on freeways and bridges. The design implemented in this paper attempts to capture the license plate by using three phase detection process that helps to increase the level of security and contribute in making a sustainable city.

This paper presents a distinctive approach to detect the license plate of the vehicles using the various image processing techniques such as dilation, grey-scale conversion, edge processing, etc. and finding the region of interest of the segmented image to capture the license plate of the vehicles.

In this chapter, drone and vision camera technology have been combined for monitoring the crop product quality. Three vegetable crops such as tomato, cauliflower, and eggplant are considered for quality monitoring; hence, image datasets are collected for those vegetables only. The proposed method classified the vegetables into two classes as rotten and nonrotten products so the images were collected for rotten and nonrotten products. Three different features information such as chromatic features, contour features, and texture features have been extracted from the dataset and further used to train a Gaussian kernel support vector machine algorithm for identifying the product quality. The system utilized multiple features such as chromatic, contour, and texture features in classifier training which enhances the accuracy and robustness of the system. Chromatic features were utilized for detecting the crop while other features such as contour and texture features were utilized for further classifier building to identify the crop product quality. The performance of the system is evaluated based on the true positive rate, false discovery rate, positive predictive value, and accuracy. The proposed system identified good and bad products with a 97.9% of true positive rate, 2.43 % of false discovery rate, 97.73% positive predictive value, and 95.4% of accuracy. The achieved results concluded that the results are lucrative and the proposed system is efficient in agriculture product quality monitoring.

The purpose of this study is to develop a cost-effective autonomous wheelchair robot navigation method that assists the aging population.

Navigation in outdoor environments is still a challenging task for an autonomous mobile robot because of the highly unstructured and different characteristics of outdoor environments. This study examines a complete vision guided real-time approach for robot navigation in urban roads based on drivable road area detection by using deep learning. During navigation, the camera takes a snapshot of the road, and the captured image is then converted into an illuminant invariant image. Subsequently, a deep belief neural network considers this image as an input. It extracts additional discriminative abstract features by using general purpose learning procedure for detection. During obstacle avoidance, the robot measures the distance from the obstacle position by using estimated parameters of the calibrated camera, and it performs navigation by avoiding obstacles.

The developed method is implemented on a wheelchair robot, and it is verified by navigating the wheelchair robot on different types of urban curve roads. Navigation in real environments indicates that the wheelchair robot can move safely from one place to another. The navigation performance of the developed method and a comparison with laser range finder (LRF)-based methods were demonstrated through experiments.

This study develops a cost-effective navigation method by using a single camera. Additionally, it utilizes the advantages of deep learning techniques for robust classification of the drivable road area. It performs better in terms of navigation when compared to LRF-based methods in LRF-denied environments.

Spray curing has become the preferred curing method for most cement concrete members because of its lower cost and sound effect. However, the spray curing quality of members is vulnerable to random variation environment factors and anthropogenic interferences. This paper aims to introduce the machine learning algorithm into the spray curing system to optimize its control method to improve the spray curing quality of members.

The critical parameters affecting the spray curing quality of members were collected through experiments, such as the temperature and humidity of the member's surface, the temperature, humidity and wind speed of the environment. The C4.5 algorithm was used as a weak classifier algorithm, and the AdaBoost.M1 algorithm was used to cascade multiple weak classifiers to form a robust classifier according to the collected data.

The results showed that the model constructed by the AdaBoost.M1 algorithm had achieved higher accuracy and robustness among the two algorithms. Based on the classification model built by the AdaBoost.M1 algorithm, the spray curing system can cause automatic decision-making spray switching according to the member's real-time curing state and environment.

With the classification model constructed by the AdaBoost.M1 algorithm, the spray curing system can overcome the disadvantages that external factors greatly influence the current control method of the spray curing system, and the intelligent control of the spray curing system was realized to a certain extent. This paper provides a reference for applying machine learning algorithms in the intellectual transformation of bridge construction equipment.

Facility management gained profound importance due to the increasing complexity of different systems and the cost of operation and maintenance. However, due to the increasing complexity of different systems, facility managers may suffer from a lack of information. The purpose of this paper is to propose a new facility management approach that links segmented assets to the vital data required for managing facilities.

Automatic point cloud segmentation is one of the most crucial processes required for modelling building facilities. In this research, laser scanning is used for point cloud acquisition. The research utilises region growing algorithm, colour-based region-growing algorithm and Euclidean cluster algorithm.

A case study is worked out to test the accuracy of the considered point cloud segmentation algorithms utilising metrics precision, recall and F-score. The results indicate that Euclidean cluster extraction and region growing algorithm revealed high accuracy for segmentation.

The research presents a comparative approach for selecting the most appropriate segmentation approach required for accurate modelling. As such, the segmented assets can be linked easily with the data required for facility management.

The paper aims to focus on a vision-based approach to advance the automated process of the manufacturing of an Airbus A350’s pressure bulkhead. The setup enables automated deformation and draping of a fiber textile on a form-variable end-effector.

The proposed method uses the information of infrared (IR) and color-based images in Red, Green and Blue (RGB) representative format, as well as depth measurements to identify the wrinkles and boundary edge of semi-finished dry fiber products on the double-curved surface of a flexible modular gripper used for laying the fabric. The technique implements a simple and practical image processing solution using a sequence of pixel-wise binary masks on an industrial scale setup; it bridges the gap between laboratory experiments and real-world execution, thereby demonstrating practical and applied research.

The efficacy of the technique is demonstrated via experiments in the presented work. The two objectives as follows boundary edge detection and wrinkle detection are accomplished in real time in an industrial setup.

During the draping process, tensions developed in the fibers of the textile cause wrinkles on the surface, which are highly detrimental to the production process, material quality and strength. The proposed method automates the identification and detection of the wrinkles and the textile on the gripper surface. The proposed work aids in alleviating the problems caused by these wrinkles and helps in quality control in the production process.

Fire is a common disaster. Even though simple sensors such as those detecting smoke or heat are popularly employed, they require close proximity to fire. In order to obtain more reliable and more complete information, fire detection by vision sensing has recently acquired increasing attention. In the vision-based fire sensing, colour is usually used as an important cue for flame detection. However, considering there are still a large number of black-and-white (B/W) CCTV cameras installed for security purposes, a technique that can detect flame reliably in grey-scale images will be useful to protect human lives and property from the fire disaster. The paper aims to discuss these issues.

This article describes the automatic detection of fire flames in the grey-scale image sequences by a two-level image processing scheme: pixel-level and frame-level. In pixel-level processing, an evaluation function is devised to extract pixels that possibly belong to the flame region, particularly to its boundaries. Extracted fire pixel candidates are verified in frame-level processing by monitoring their distribution variations in sequential images. A circle is fitted to the candidate pixels in each image for efficient monitoring, and the presence of flame is reasoned when the position and size of the circle increase with high fluctuations.

Experimental results show that the proposed method can detect flame quite reliably using the intensity information and its temporal variations in grey-scale image sequences.

This paper presents a novel technique of vision-based flame detection. Unlike most existing techniques, the proposed technique is based on the grey-scale images of a B/W camera. To the best of the author's knowledge, it may be the first of its kind developed for general application to indoor and outdoor scenes.

This paper investigates interference mitigation and read rate improvement by using novel power control and graph‐based scheduling schemes for radio frequency identification (RFID) systems.

The first method is a distributed power control (DPC) scheme proposed as an alternative to listen‐before‐talk (LBT) for RFID systems specified under CEPT regulations. The DPC algorithm employs reader transmission power as the system control variable to achieve a desired read range and read rate without causing unwanted interference. The second approach is graph‐based scheduling, which uses a graph coloring‐based approach to temporally separate readers with overlapping interrogation zones. The scheduling of the timeslots is carried out so as to offer better efficiency for each reader.

This paper shows that power control, graph theory, collision probability analysis along with timeslot scheduling schemes can be widely adapted to solve general RFID problems. The study shows that selection of timeslot allocation schemes should be carried out after carefully analysing the process/workflow in the application domain. While fair scheduling schemes can be applicable to stable manufacturing environments, event‐triggered scheduling schemes are more effective in fairly chaotic environments.

The study shows that the proposed interference mitigation and read rate improvement techniques can be generalized to assist in design, development, and implementation of a variety of RFID‐based systems, ranging from supply chain level operations to shop floor control. The proposed techniques improve not only the reliability of RFID systems but, more importantly, improve business processes that rely on RFID data.

The purpose of this paper is to build a classification system which mimics the perceptual ability of human vision, in gathering knowledge about the structure, content and the surrounding environment of a real-world natural scene, at a quick glance accurately. This paper proposes a set of novel features to determine the gist of a given scene based on dominant color, dominant direction, openness and roughness features.

The classification system is designed at two different levels. At the first level, a set of low level features are extracted for each semantic feature. At the second level the extracted features are subjected to the process of feature evaluation, based on inter-class and intra-class distances. The most discriminating features are retained and used for training the support vector machine (SVM) classifier for two different data sets.

Accuracy of the proposed system has been evaluated on two data sets: the well-known Oliva-Torralba data set and the customized image data set comprising of high-resolution images of natural landscapes. The experimentation on these two data sets with the proposed novel feature set and SVM classifier has provided 92.68 percent average classification accuracy, using ten-fold cross validation approach. The set of proposed features efficiently represent visual information and are therefore capable of narrowing the semantic gap between low-level image representation and high-level human perception.

The method presented in this paper represents a new approach for extracting low-level features of reduced dimensionality that is able to model human perception for the task of scene classification. The methods of mapping primitive features to high-level features are intuitive to the user and are capable of reducing the semantic gap. The proposed feature evaluation technique is general and can be applied across any domain.