Implementing an innovative theoretical proposal with optimization in the time of computational processing involving image vectorization techniques

Factor analysis-based approach for early uptake automatic quantification of breast cancer by 18F-FDG PET images sequence

In this paper, we proposed a fast and efficient high dynamic range reconstruction model based on brightness compensation and multi-source bidirectional similarity (MBDS). By introducing the exposure value detecting function, the dynamic image sequence with different exposure values is divided into the high-exposure images, the low-exposure images and the reasonable exposure images. Considering that the brightness dynamic range is an important factor which greatly influences the image visual effect and its rich information contents, we improve the image brightness by the brightness enhancement algorithm for the low-exposure images. By combining the exposure value detecting and the brightness enhancement, it could improve the time efficiency and the block matching accuracy for the high dynamic range reconstruction algorithm based on MBDS, and also the artifacts and shake phenomenon produced in the reconstruction process could be reduced. Experimental results demonstrate that the proposed model could achieve better exposure compensation, which obtains more reasonable image brightness and much richer image details information.

This paper presents a fast and reliable algorithm for fingerprint verification. Our proposed fingerprint verification algorithm is based on image-based fingerprint matching. The improved orientation feature vector of two fingerprints has been compared to compute the similarities at a given threshold. Fingerprint image has been aligned by rotating through an angle before feature vector is computed and matched. Row and Column variance feature vector of orientation image will be employed. The algorithm has been tested on the FVC2002 Databases. The performance of algorithm is measured in terms of GAR and FAR. At a threshold level of 1.1 % and at 5.7% FAR the GAR observed is 97.83%. The improved Feature vector will lower imposter acceptance rate at reasonable GAR and hence yields better GAR at lower FAR. The proposed algorithm is computationally very efficient and can be implemented on Real-Time Systems.

Recently, there has been increased interest in imaging the coronary vein anatomy to guide interventional cardiovascular procedures such as cardiac resynchronization therapy (CRT), a device therapy for congestive heart failure (CHF). With CRT the lateral wall of the left ventricle is electrically paced using a transvenous coronary sinus lead or surgically placed epicardial lead. Proper transvenous lead placement is facilitated by the knowledge of the coronary vein anatomy. Cardiovascular MR (CMR) has the potential to image the coronary veins. In this study we propose and test CMR techniques and protocols for imaging the coronary venous anatomy. Three aspects of design of imaging sequence were studied: magnetization preparation schemes (T(2) preparation and magnetization transfer), imaging sequences (gradient-echo (GRE) and steady-state free precession (SSFP)), and imaging time during the cardiac cycle. Numerical and in vivo studies both in healthy and CHF subjects were performed to optimize and demonstrate the utility of CMR for coronary vein imaging. Magnetization transfer was superior to T(2) preparation for contrast enhancement. Both GRE and SSFP were viable imaging sequences, although GRE provided more robust results with better contrast. Imaging during the end-systolic quiescent period was preferable as it coincided with the maximum size of the coronary veins.

The analysis of a class of complex images has been simplified by extracting the edge-dominated features before matching the sequential images. The consecutive images are then registered by a frequency-domain technique, specifically by a combination of two-dimensional power spectrum and cepstrum techniques to correct for rotational and translational shifts, respectively. The cepstrum technique is found to be more accurate for correction of a translational shift than are the commonly used phase-correlation techniques and spatial-domain-correlation techniques, particularly for noisy and nonuniformly featured sequential images. The change in sequential images is expressed quantitatively in terms of the mean and the variance of the computed two-dimensional histogram representing the difference of two consecutive images. Such quantitative measures of change in sequential images have been applied to a class of complex medical images, namely, retinal (fundus) images, to provide a diagnostic measure for early detection of glaucoma. However, the general procedure of using feature-extraction techniques first and then registering and analyzing images by using power-spectrum and two-dimensional cepstrum techniques provides an unambiguous, accurate, and fast technique for the analysis of a broad range of sequential complex images.

We investigated an approach to reconstructing high‐resolution images from dynamic image sequences using local spectral analysis. High‐resolution reconstruction from linearly shifted multiple static image frames has previously been studied, in which the aliasing relationship between the spectrum of the original signal and the DFTs of shifted and sampled signals is exploited. In the high‐resolution reconstruction of dynamic image sequences, difficulties occur as a result of nonuniform shifts in the frames. We use loca spectral analysis to achieve high‐resolution reconstruction by overlapped block decomposition and motion compensation. For each block image in a reference frame in the sequence, motion estimation and subpixel registrations are performed against adjacent frames. High resolution reconstruction is performed on such motion‐compensated block‐image sequences. For some blocks containing motion boundaries, high resolution reconstruction is difficult; a new scene emerges or disappears producing inconsistent information. An interpolation technique is used in such blocks to generate the number of pixels consistent with other high‐resolution blocks. The flower‐garden image sequence is used for the computer simulations. The quality of the restored images are very encouraging; the aliasing effects in the original frames are significantly reduced and sharper edges are produced. The overall procedure to generate such higher‐resolution images from a dynamic image sequence is described in detail. The result can be applied to various image sequence restoration applications including up‐conversion of conventional video signals.©1994 John Wiley & Sons Inc

The entropy sequence of the output image, gotten from the original gray image by pulse-coupled neural network (PCNN), as feature vector of the gray image, can be used as a unique feature expression of gray image, which has been proved by our experiment, therefore, in this paper, it is used in the image classification, and the mean square error (MSE) between the feature vector of the input image and standard feature vector is used to judge the input image belong to which kind of image groups. At the same time, the results of our experiment show that this method is strongly flexible to resist noises and greatly robust to recognize image, if the tested images in our experiment are disturbed with Gaussian noise, impulse noise or both of this

Eye fatigue while driving can cause drivers to be drowsy and less alert, which can potentially increase the risk of an accident.Existing data shows that the number of accidents in the world is increasing from year to year.One of the most common causes of accidents is fatigue and the leading cause of death is car accidents.Therefore, efforts are needed to reduce accidents due to fatigue.To overcome this, in this study, a system was developed to detect driver eye fatigue using the Convolutional Neural Network method with varying image sizes as input.The dataset consists of 1289 facial images that contain the eyes and is divided into 614 drowsiness eyes and 675 non-drowsiness eyes.In dealing with variations in image size, scaling was carried out using five interpolation methods, namely nearest-neighbor, bilinear, bicubic, inter-area, and lanczos4.The performance of the sleepy eye detection model will be evaluated based on accuracy and processing time.The results show that the image size of 64 64 with bilinear interpolation and 96 96 with inter-area interpolation gives the highest accuracy of 99%.Based on processing time, resizing the image to 8 8 size by using bilinear, bicubic, inter-area, and lanczos4 interpolation, results in the fastest processing time and high accuracy of 94% -95%.The difference in accuracy with other image sizes is only 5%, with processing time for other size images up to 200 times longer than processing time for 8 8 image sizes.

This paper investigated the effects of critical-point drying (CPD) and hexamethyldisilazane (HMDS) sample preparation techniques for cervical cells on field emission scanning electron microscopy and energy dispersive X-ray (FE-SEM/EDX). We investigated the visualization of cervical cell image and elemental distribution on the cervical cell for two techniques of sample preparation. Using FE-SEM/EDX, the cervical cell images are captured and the cell element compositions are extracted for both sample preparation techniques. Cervical cell image quality, elemental composition, and processing time are considered for comparison of performances. Qualitatively, FE-SEM image based on HMDS preparation technique has better image quality than CPD technique in terms of degree of spread cell on the specimen and morphologic signs of cell deteriorations (i.e., existence of plate and pellet drying artifacts and membrane blebs). Quantitatively, with mapping and line scanning EDX analysis, carbon and oxygen element compositions in HMDS technique were higher than the CPD technique in terms of weight percentages. The HMDS technique has shorter processing time than the CPD technique. The results indicate that FE-SEM imaging, elemental composition, and processing time for sample preparation with the HMDS technique were better than CPD technique for cervical cell preparation technique for developing computer-aided screening system.

Visualizing children’s stories is an important task to support distance learning and education during the current COVID-19 pandemic outbreak. This task can be regarded as a rapid transition to overcome any significant disruption to the provision of visual learning for children. However, such a task needs a deep understanding of the characters and events being involved in a story text. It also includes the distinction between common actions and complex ones. The distinction helps in visualizing the objects and their relationships in the scene in a consistent and coherent way. In addition, this process requires natural language processing and understanding as well as a dynamic sequence of images rather than presenting a single image. Specifically, handling datasets with high semantic complexity as in the case of children’s stories needs detailed image sequence to support the visualization task. Some recent works have utilized Generative Adversarial Networks (GANs) which show some promising upshots to tackle this challenge. Despite these works success to generate images for isolated sentences, they fail to generate images that are well consistent to the description and coherent to the whole story. To address this issue, we propose a method to generate single images and sequences of images for common and non-common actions, respectively. This yields different sequences of images which we evaluate based on a coherency metric. Our obtained results show the effectiveness of the proposed method.KeywordsScene generationStory visualizationGANStory understandingLanguage learning

Image processing technology is often combined with deep learning to be applied in various recognition and classification tasks. However, most existing image processing technologies struggle to balance both accuracy and efficiency. To address this issue, this paper proposes an efficient image processing model that embeds Nesterov's accelerated gradient optimization imprecise block coordinate descent algorithm into a super-resolution reconstruction framework. The model first divides the input image into blocks and uses the Mean Shift algorithm to cluster the image blocks to reduce computational complexity. Subsequently, the Nesterov accelerated gradient algorithm was used to accelerate the descent process of the imprecise block coordinate algorithm, and finally the image resolution was improved through super-resolution reconstruction. The experiment is based on the DIV2K dataset, with the alternating direction method of multipliers, stochastic gradient descent, and Super-Resolution Convolutional Neural Network models as comparison baselines, and it is validated in a hardware environment equipped with an AMD Radeon RX 6800 XT GPU. The results show that the average Peak Signal to Noise Ratio and Structural Similarity Index of the proposed algorithm are 33.43dB and 0.916, respectively. The average processing time for a single image is 16ms, and the overall similarity of the images is 90.1%. The minimum image processing time for the proposed image processing model is 36ms, and the processing time for a single image does not exceed 40ms. These results demonstrate that the proposed model not only ensures high accuracy but also meets the efficiency requirements, allowing for effective restoration and optimization of target images. The proposed method offers new insights for image processing and contributes to the optimization of various techniques based on image processing.

Image vectorization is one of the primary means of creating vector graphics. The quality of a vectorized image depends crucially on extracting accurate features from input raster images. However, correct object edges can be difficult to detect when color gradients are weak. We present an image vectorization technique that operates on a color image augmented with a depth map and uses both color and depth edges to define vectorized paths. We output a vectorized result as a diffusion curve image. The information extracted from the depth map allows us more flexibility in the manipulation of the diffusion curves, in particular permitting high-level object segmentation. Our experimental results demonstrate that this method achieves high reconstruction quality and provides greater control in the organization and editing of vectorized images than existing work based on diffusion curves.

Current image vectorization techniques mainly deal with images with simple and plain colors. For full-color photographs, many difficulties still exist in object segmentation, feature line extraction, and color distribution reconstruction, etc. In this paper, we propose a high-efficiency image vectorization method based on importance sampling and triangulation. A set of blue-noise sampling points is first generated on the image plane by an improved error-diffusion sampling method. The point set well preserves the features in the image. Then after triangulation on this point set, color information can be recorded on the mesh vertices to form a vector image. After certain image editing, e.g. scaling or transforming, the whole image can be reconstructed by color interpolating inside each triangle. Experiments show that the method has high performing efficiency and abilities in feature-preserving. It will bring benefits to many applications, e.g. image compressing, editing, transmitting and resolution enhancement.

To assess the effect of standard filtered back projection (FBP) and iterative reconstruction (IR) methods on CBCT image noise and processing time (PT), acquired with various acquisition parameters with and without metal artefact reduction (MAR). CBCT scans using the Midmark EIOS unit of a human mandible embedded in soft tissue equivalent material with and without the presence of an implant at mandibular first molar region were acquired at various acquisition settings (milliamperages [4mA-14mA], FOV [5 × 5, 6 × 8, 9 × 10 cm], and resolutions [low, standard, high] and reconstructed using standard FBP and IR, and with and without MAR. The processing time was recorded for each reconstruction. ImageJ was used to analyze specific axial images. Radial transaxial fiducial lines were created relative to the implant site. Standard deviations of the gray density values (image noise) were calculated at fixed distances on the fiducial lines on the buccal and lingual aspects at specific axial levels, and mean values for FBP and IR were compared using paired t-tests. Significance was defined as p < 0.05. The overall mean for image noise (± SD) for FBP was 198.65 ± 55.58 and 99.84 ± 16.28 for IR. IR significantly decreased image noise compared to FBP at all acquisition parameters (p < 0.05). Noise reduction among different scanning protocols ranged between 29.7% (5 × 5 cm FOV) and 58.1% (5mA). IR increased processing time by an average of 35.1 s. IR significantly reduces CBCT image noise compared to standard FBP without substantially increasing processing time.

The focus of this work is the development of an efficient adaptive algorithm for source separation from a noisy image sequence with interferences when the underlying source signals are unknown. The sources to be extracted are the underlying activation signals who are collectively responsible for the intensity fluctuations of the pixels. This technique uses the concept of temporal Independent Component Analysis (tICA) which does not rely on source signal information for its unmixing process. Each pixel of an image is taken as a sensor which has its own intensity fluctuation pattern. Thus the whole image is a collection of two dimensional signal mixtures which can be used to separate both super and sub Gaussian sources through an adaption process. This work demonstrates the viability of utilizing the concept of subspace separation and tICA for removal of visual interferences from dynamic image sequences when extracting underlying source signals. Two methods are proposed: The first of which identifies the interference at the ICA output level while the second method removes the interference at the subspace separation stage. The proposed technique has a potential use in a wide array of applications such as computer vision, surveillance bio-medicine, etc.