Proposed Image Processing Research Articles

Image Dehazing Enhancement Strategy Based on Polarization Detection of Space Targets

In view of the fact that the technology of polarization detection performs better at identifying targets through clouds and fog, the recognition ability of the space target detection system under haze conditions will be improved by applying the technology. However, due to the low ambient brightness and limited target radiation information during space target detection, the polarization information of space target is seriously lost, and the advantages of polarization detection technology in identifying targets through clouds and fog cannot be effectively exerted under the condition of haze detection. In order to solve the above problem, a dehazing enhancement strategy specifically applied to polarization images of space targets is proposed. Firstly, a hybrid multi-channel interpolation method based on regional correlation analysis is proposed to improve the calculation accuracy of polarization information during preprocessing. Secondly, an image processing method based on full polarization information inversion is proposed to obtain the degree of polarization of the image after inversion and the intensity of the image after dehazing. Finally, the image fusion method based on discrete cosine transform is used to obtain the dehazing polarization fusion enhancement image. The effectiveness of the proposed image processing strategy is verified by carrying out simulated and real space target detection experiments. Compared with other methods, by using the proposed image processing strategy, the quality of the polarization images of space targets obtained under the haze condition is significantly improved. Our research results have important practical implications for promoting the wide application of polarization detection technology in the field of space target detection.

A centralized frost detection and estimation scheme for Internet-connected domestic refrigerators

Frost accumulation on heat exchange units is a significant problem in refrigeration systems, adversely affecting their operating performance and thereby leading to increased power consumption. Therefore, timely detection and accurate quantification of frost are crucial for effective defrosting strategies. This study presents a novel centralized cloud-based IoT scheme for frost detection and thickness estimation. The image processing is performed on the cloud server to process evaporator coil images for frost thickness quantification. Experiments were conducted on a domestic refrigerator to assess the effectiveness of the proposed image-processing approach and determine latency and processing time. The presented scheme effectively quantifies frost thickness on the evaporator in the 1–5 mm range with a 10.8% error margin. The total inference time, which includes image acquisition, pre-processing, transmission latency, and frost thickness estimation, is approximately 5.15 seconds. The results demonstrate that the proposed image processing method performs comparably to conventional sensors and similar image processing techniques. Moreover, the centralized cloud-based IoT architecture presented effectively meets the scalability demands of consumer refrigerators.

For a clinical application of optical triangulation to assess respiratory rate using an RGB camera and a line laser

This paper presents a non-contact and unrestrained respiration monitoring system based on the optical triangulation technique. The proposed system consists of a red-green-blue (RGB) camera and a line laser installed to face the frontal thorax of a human body. The underlying idea of the work is that the camera and line laser are mounted in opposite directions, unlike other research. By applying the proposed image processing algorithm to the camera image, laser coordinates are extracted and converted to world coordinates using the optical triangulation method. These converted world coordinates represent the height of the thorax of a person. The respiratory rate is measured by analyzing changes of the thorax surface depth. To verify system performance, the camera and the line laser are installed on the head and foot sides of a bed, respectively, facing toward the center of the bed. Twenty healthy volunteers were enrolled and underwent measurement for 100s. Evaluation results show that the optical triangulation-based image processing method demonstrates non-inferior performance to a commercial patient monitoring system with a root-mean-squared error of 0.30rpm and a maximum error of 1rpm (p>0.05\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$p > 0.05$$\\end{document}), which implies the proposed non-contact system can be a useful alternative to the conventional healthcare method.

Bornean Orangutan Nest Classification using Image Enhancement with Convolutional Neural Network and Kernel Multi Support Vector Machine Classifier

Preserving wildlife habitats is crucial in mitigating climate change. Species like orangutans and monkeys contribute to fruiting and planting in forests. The World Wide Fund Sabah Malaysia faces challenges in manually identifying and classifying orangutan nests for studying their behaviour and conserving their habitats. To address this, we propose automating the classification of captured images using machine learning algorithms. This research involves three key components: image processing, feature extraction, and image classification. Our proposed image processing includes several steps, such as image pre-processing and enhancement techniques like local contrast enhancement, sharpening, intensity adjustment, histogram equalization, and colour thresholding. We applied four different Convolutional Neural Networks (CNNs) to extract and identify orangutan nests’ features. Subsequently, we utilize Support Vector Machine (SVM) for image classification. The results reveal that the Inception Residual Network Version 2 (ResNet-v2) achieves the best performance. This architecture is then combined with a kernel SVM to classify Bornean orangutan nests. Our approach demonstrates impressive results, boasting an accuracy of 96.60%, an F1-score of 96.60%, a precision of 96.59%, and a recall of 96.58%. These metrics underscore the high accuracy and effectiveness of our proposed methodology for classifying Bornean orangutan nests. By reducing the need for extensive human intervention in image analysis, our method presents a valuable tool for conservationists and researchers committed to studying and safeguarding these endangered orangutans and their habitats. In future work, we aim to develop orangutan nest detector, contributing to wildlife conservation research.

Computer vision–based surface defect identification method for weld images

Welding is the main connection type used in steel structure bridges. As the weld quality significantly affects the performance of a structure, weld defect detection is crucial in the performance evaluation of bridges. With the continuous development of computer vision (CV) techniques and image acquisition equipment, CV-based methods are being increasingly applied to weld defect detection. However, accurate identification of welding defects from several images is challenging. This paper proposes a CV-based method for detecting and quantifying surface porosity defects in welds. CV techniques were used to identify and extract weld defects, and the pixel-level sizes of the weld defects were obtained. The actual sizes of the weld defects were calculated based on a conversion between the image size and actual size. The results show that the proposed image processing method has low complexity and high accuracy in detecting weld defects, providing accurate measurement of the defect size. It is a fast and non-contact detection method for weld defect detection.

Smart Agriculture Applications and Data Analysis with Industry 4.0

The world’s food demand is predicted to exceed its food supply significantly in the coming periods. Many physical factors, such as climate change in access to water, change the production amount and food prices. Efficient resource utilization can be achieved by developing smart agricultural applications using the IoT and image processing techniques. For this purpose, in the proposed study, a smart agriculture application was developed using IoT technology with image processing. In the proposed image processing algorithm, the crops’ images were processed, weight estimates were made, and their numbers were taken. The data we collect instantly can be uploaded to the cloud, and the data collected on the cloud side can be analyzed over the mobile application. Thanks to data analysis with the mobile application we have developed, we can see the current status of the crops and make instant decisions, and it also plays a role in determining market prices. With all these studies, it has been seen that agriculture can be digitalized with the technologies we use, and thus, productivity in agriculture can increase significantly.

MA‐ResUNet: Multi‐attention optic cup and optic disc segmentation based on improved U‐Net

AbstractGlaucoma poses a significant threat to vision, capable of causing irreversible damage and, in severe instances, leading to permanent blindness. Accurate optic cup (OC) and optic disc (OD) segmentation are essential in glaucoma screening. In this study, a novel OC and OD segmentation approach is proposed. Based on U‐Net, it is optimized by introducing cardinality dimensions. Moreover, attention gates are implemented to reinforce salient features while suppressing irrelevant information. Additionally, a convolutional block attention module (CBAM) is integrated into the decoder segment. This fusion hones in on effective information in both channel and spatial dimensions. Meanwhile, an image processing procedure is proposed for image normalization and enhancement. All of these increase the accuracy of the model. This model is evaluated on the ORIGA and REFUGE datasets, demonstrating the model's superiority in OC and OD segmentation compared to the state‐of‐the‐art methods. Additionally, after the proposed image processing, cup‐to‐disc ratio (CDR) prediction on a batch of 155 in‐house fundus images yields an absolute CDR error of 0.099, which is reduced by 0.04 compared to the case where only conventional processing was performed.

P‐236: Late‐News Poster: Hard defect detection and classification for display panel products

Hard defect of display panel such as crack or chipping is one of the most critical defects that have fatal impacts on the product. Conventional deep‐learning‐based methods have difficulties in equipment control in practical point of view due to the vastness of training data and numerous parameters. Consequently, more simple and robust machine‐learning models jointly with traditional image processing algorithms with small number of parameters are increasingly required for the maintenance of inspection equipment. To resolve the issue, we proposed a novel image processing and decision tree classification algorithm that can be operated using smaller number of parameters than deep‐learning method. Firstly, panel image is fed into proposed image processing algorithm and converted into feature map to highlight the hard defect. After binary thresholding of the image, remaining pixels are specified as bounding box. Subsequently, features are extracted in the bounding box followed by classification into normal/defective panel. Additional RANSAC‐based outer edge inspection algorithm is processed. We conducted an experimental study using manufacturing data, and showed successful performance in hard defect detection and classification.

Computer-Vision- and Deep-Learning-Based Determination of Flow Regimes, Void Fraction, and Resistance Sensor Data in Microchannel Flow Boiling.

The aim of this article is to introduce a novel approach to identifying flow regimes and void fractions in microchannel flow boiling, which is based on binary image segmentation using digital image processing and deep learning. The proposed image processing pipeline uses adaptive thresholding, blurring, gamma correction, contour detection, and histogram comparison to separate vapor from liquid areas, while the deep learning method uses a customized version of a convolutional neural network (CNN) called U-net to extract meaningful features from video frames. Both approaches enabled the automatic detection of flow boiling conditions, such as bubbly, slug, and annular flow, as well as automatic void fraction calculation. Especially CNN demonstrated its ability to deliver fast and dependable results, presenting an appealing substitute to manual feature extraction. The U-net-based CNN was able to segment flow boiling images with a Dice score of 99.1% and classify the above flow regimes with an overall classification accuracy of 91%. In addition, the neural network was able to predict resistance sensor readings from image data and assign them to a flow state with a mean squared error (MSE) < 10-6.

Image processing algorithm in the spatial navigation system for the visually impaired

The algorithm for processing the image generated by the webcam of the spatial navigation system for the blind and visually impaired is proposed. Image processing includes selecting and identifying objects located in four sectors of the frame, shifting the center of the frame to the selected objects and determining the angular coordinates of the objects' location, forming an indication for turning the user's head towards the selected objects, measuring the distance to the object using LiDar, highlighting foreground objects, determining the speed of its movement and ensuring safe navigation visually impaired in accordance with the information voiced by the speech synthesizer. The proposed image processing algorithm is designed to form information support for the blind and visually impaired in spatial navigation systems.

An automatic system for recognizing fly courtship patterns via an image processing method

Fruit fly courtship behaviors composed of a series of actions have always been an important model for behavioral research. While most related studies have focused only on total courtship behaviors, specific courtship elements have often been underestimated. Identifying these courtship element details is extremely labor intensive and would largely benefit from an automatic recognition system. To address this issue, in this study, we established a vision-based fly courtship behavior recognition system. The system based on the proposed image processing methods can precisely distinguish body parts such as the head, thorax, and abdomen and automatically recognize specific courtship elements, including orientation, singing, attempted copulation, copulation and tapping, which was not detectable in previous studies. This system, which has high identity tracking accuracy (99.99%) and high behavioral element recognition rates (> 97.35%), can ensure correct identification even when flies completely overlap. Using this newly developed system, we investigated the total courtship time, and proportion, and transition of courtship elements in flies across different ages and found that male flies adjusted their courtship strategy in response to their physical condition. We also identified differences in courtship patterns between males with and without successful copulation. Our study therefore demonstrated how image processing methods can be applied to automatically recognize complex animal behaviors. The newly developed system will largely help us investigate the details of fly courtship in future research.

Automating RTGC with PID Control: Utilizing Camera-Based Image Processing and Object Detection

The high traffic in the container yard requires effective management, and one method to address this is through automation. The Rubber Tyred Gantry Crane (RTGC) plays a crucial role in container yards. Automating the RTGC involves determining the container’s location and sway angle to provide feedback for the control system. The advancements in computer vision technology offer a unique solution to tackle these challenges. This paper proposes image processing as a means to sense the sway angle and MobileNet SSD to detect the container’s location. The proposed method yields accurate measurement results and is integrated with optimized PID-PD for position and sway angle control in RTGC. The effectiveness of the proposed method is demonstrated through successful performance in both simulation and experiments conducted on a laboratory-scale RTGC prototype.

A Novel Image Processing Method for Obtaining an Accurate Three-Dimensional Profile of Red Blood Cells in Digital Holographic Microscopy.

Recently, research on disease diagnosis using red blood cells (RBCs) has been active due to the advantage that it is possible to diagnose many diseases with a drop of blood in a short time. Representatively, there are disease diagnosis technologies that utilize deep learning techniques and digital holographic microscope (DHM) techniques. However, three-dimensional (3D) profile obtained by DHM has a problem of random noise caused by the overlapping DC spectrum and sideband in the Fourier domain, which has the probability of misjudging diseases in deep learning technology. To reduce random noise and obtain a more accurate 3D profile, in this paper, we propose a novel image processing method which randomly selects the center of the high-frequency sideband (RaCoHS) in the Fourier domain. This proposed algorithm has the advantage of filtering while using only recorded hologram information to maintain high-frequency information. We compared and analyzed the conventional filtering method and the general image processing method to verify the effectiveness of the proposed method. In addition, the proposed image processing algorithm can be applied to all digital holography technologies including DHM, and in particular, it is expected to have a great effect on the accuracy of disease diagnosis technologies using DHM.

Breast cancer classification based on convolutional neural network and image fusion approaches using ultrasound images

Deep learning and image processing are used to classify and segment breast tumor images, specifically in ultrasound (US) modalities, to support clinical decisions and improve healthcare quality. However, directly using US images can be challenging due to noise and diverse imaging modalities. In this study, we developed a three-step image processing scheme involving speckle noise filtering using a block-matching three-dimensional filtering technique, region of interest highlighting, and RGB fusion. This method enhances the generalization of deep-learning models and achieves better performance. We used a deep learning model (VGG19) to perform transfer learning on three datasets: BUSI (780 images), Dataset B (162 images), and KAIMRC (5693 images). When tested on the BUSI and KAIMRC datasets using a fivefold cross-validation mechanism, the model with the proposed preprocessing step performed better than without preprocessing for each dataset. The proposed image processing approach improves the performance of the breast cancer deep learning classification model. Multiple diverse datasets (private and public) were used to generalize the model for clinical application.

Early Detection of The Glaucoma and Other Intra-Ocular Pressure Elevation Diseases Using Hardware Efficient Machine Learning Approach

Nowadays, Glaucoma is one of the chronical diseases that entirely make the human eyes into the blindness. This disease is a consequence of an accumulation of aqueous humor in the eye due to a defect of its drainage system. This condition progressively elevates the intra-ocular pressure (IOP), affecting the optic nerve and resulting in permanent blindness if left untreated. In early stages, the glaucoma may be an asymptomatic. Hence, the proposed method is designed to detect the early stage of the glaucoma. This can be done by measuring the cup to disk ratio. For that, the proposed image processing algorithm is constrained with the three basic steps such as preprocessing, feature extraction and classification. In classification stage, we employ the SVM classifier to classify the normal and glaucoma images. The method is found to be efficient in hardware implementation when compared to other methods. The overall implementation will be held in the Matlab supporting environment.

Time-Frequency Domain Monitoring Method for the Fault of HTS HVDC Systems Based on AI Classifiers

Since high-temperature superconductor (HTS) based HVDC systems are affected by temperature and cooling system performance, noise, the monitoring system which can classify the temporal disturbance and anomalies must be developed to ensure stable system operation. To solve this, it needs anomaly detection that accounts for system sensitivity and is able to continuously manage the response to disturbances. In this paper, we propose an integrated solution for fault severity diagnosis and anomaly detection colorred with AI based reflectometry. Anomalies such as the quench phenomenon can be detected through the anomaly score based on the reconstruction error calculated through the proposed autoencoder(AE) method. After anomaly detection, a fault classification algorithm based on the convolutional neural network (CNN) using a 2D image of a converted reflected signal through the proposed image processing was conducted. Based on the signal acquired from the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$7 m$</tex-math></inline-formula> length 1st generation 22.9 kV/50 MVA HTS cable, PSCAD simulation was utilized to construct a long-distance line model and verified the performance of the proposed algorithm.

A fast and non-invasive imaging procedure to fight red tuna fraud

Recently, a food fraud regarding tuna samples has been detected in Europe. Some batches of tuna treated with illegal adulterants like vegetable extracts, including beetroot, were detected after several cases of food poisoning. This paper proposes a new fast and non-destructive technique based on smartphone imaging which contributes to the fight against tuna fraud. The color of different samples of three species of tuna (Thunnus thynnus, Thunnus albacares and Thunnus obesus) has been analyzed by a colorimetrically characterized smartphone. Descriptive analysis of the samples shows that different species of tuna can be grouped by their color in CIELAB space, and that samples adulterated with beetroot extracts can be detected by the proposed image processing procedure. A logistic regression model was implemented to distinguish between real and fraudulent samples, and its suitability checked by k-folding cross validation (n = 189), achieving a 0.59% false positive rate with a 10% false negative rate.

LiDAR for Object Detection in Self Driving Cars

Self-driving cars are the major invention in vehicular automation. These cars use sensors to perceive the surrounding and control them accordingly. Object detection becomes a major task in these driverless cars. In this study, utilization of LiDAR to automatically control speed, braking, and safety systems in response to sudden changes in traffic conditions is presented. The aim of this study is to do precise and quick object detection for LiDAR-based self-driving cars. This work proposes image processing to be used to identify and differentiate similar-looking objects so that carefully calculated driving decisions can be taken.

Optimum Driving of Ultrasonic Cleaner Using Impedance and FFT Analysis with Validation of Image Processing of Perforated Foils

Over the past decade, ultrasonic cleaners have been widely used in many industries. Now, this technology is finding its way into homes for vegetable, fruit, and clothes cleaning. In widely used ultrasonic cleaners, piezoelectric transducers are externally attached to the steel tank to generate ultrasonic waves inside the tank. Based on the impedance data of the piezoelectric transducers, the driving circuit was tuned to generate the required frequencies inside the cleaning tank. This paper discusses the design, development, and validation of an 800 mL tank capacity ultrasonic cleaner driven with a piezoelectric disc actuator. To achieve an optimum cleaning action without surface abrasion, several characteristics need to be considered in this complex relationship. The placement of transducers has been investigated according to the pressure distribution inside the liquid medium. The optimized ultrasonic cleaner design, along with a class-D half-bridge circuit, was developed to drive the ultrasonic transducer in the resonance frequency range. To validate the optimal design and driving frequency, the acoustic spectrum generated inside the tank was measured using a piezoelectric sensor and FFT analysis was performed. To validate the cleaning effect, a qualitative test based on aluminuim foil perforations was performed. The perforation area in the foils was quantitatively measured using image processing based on the YOLO V5 technique. The proposed image processing technique has an accuracy of 97 % in the detection of perforation areas in the aluminuim foil test.

A Scaling Factor Based Image Processing Strategy for Object Detection

Classified management of domestic garbage is conducive to controlling pollution, protecting the environment, saving resources and achieving sustainable urban development. To automate domestic garbage classification and improve classification rate and processing capacity, this paper innovatively proposes an image processing strategy to detect domestic garbage objects using domestic garbage images as a dataset and YOLOv5 network. The network is then fine-tuned to achieve object detection of domestic garbage. Experimental results show that after using the image processing strategy, mAP@.5:.95 of the first-class (4-class) and second-class (104-class) networks on the basic test set is increased from 15.4% and 10.9% to 28.4% and 18.5%, respectively. This demonstrates the feasibility and effectiveness of the proposed image processing strategy. In addition, the image processing strategies presented in this paper have the potential to be applied in the domain of video recognition, including Sign Language Translation and Lip-reading Recognition.