This retrospective study introduces a guided diffusion deep learning method to mitigate streaking artifacts in chest CT scans caused by scanner issues, physical effects, patient factors, and helical multisection techniques. Data from 763 non-contrast series (110–140 kVp, 165 mAs average, 0.5–10 mm slices) across four centers included 47,032 artifact-affected 49,609 artifact-free slices. The model was trained concatenating images with segmentation masks anatomical ROIs preserve structures...

In this study, a novel direction selection method using the two-dimensional complex discrete wavelet transform (2D-CDWT) is proposed. order to achieve arbitrary selection, directional filters are first designed. Calculation procedure of can be shown as follows: (1) The 16 sub-images generally generated from original image by 2D-CDWT without down-sampling process and 12 that correspond high-frequency components selected. (2) filtered designed filter. (3) carried out resulting images obtained....

The theorems, giving the condition of perfect translation invariance for discrete wavelet transforms, have already been proven. Based on these dual-tree complex transform, 2-dimensional packet variable-density transform and real-valued having invariance, were proposed. However, their customizability wavelets in frequency domain is limited. In this paper, also based a new type proposed, which achieves with high degree domain.

In this paper, we propose a new complex-type wavelet which call RI-Spline wavelet. We show that the satisfies admissibility condition and bi-orthogonal condition. That is, can be used for continuous

Recognizing human actions in complex scenes is a challenging problem due to background clutters, camera motion, occlusions, and illumination variations. Markov models are widely used model temporal statistical relationships among elementary for action recognition. However, traditional cannot long-range relations activities, the states of may be unstable unwanted local features. In this paper, we propose multiple-instance recognition address these issues. Our contributions twofold. First,...

As development of smart devices progresses, voice control is becoming more and important. The first step to separate noise the voice. In recent years, as data size increasing, deep learning has become a very useful tool in processing. use signals increasingly popular. However, disadvantage that output pre-processing part used multi-layer neural network suffers from oscillation when applied regression problem. this study, new wavelet been proposed improve phenomena. applies discrete transform...

In this study, we first analyze the characteristics of knocking and standing-wave by wavelet transform using ion current waveform, which is measured in a spark-ignition engine. Based on result, design wavelets to detect from waveform. We next define an evaluation parameter E, apply it detection show its effectiveness.

Purpose: Use artificial intelligence technology to identify the characteristics of covid-19 in CT images, quickly screen COVID-19 patients, achieve rapid diversion and treatment suspected reduce risk infection control spread disease.Materials methods: This article combines deep learning target detection image classification methods study images patients. By extracting analyzing features lesions different periods, a new model based on time-spatial sequence convolution is obtained. The...

Water leaks cause the ground to cave in by outflow of soil, shutting down infrastructure. So investigation water is an important issue. However it difficult detect all leak points, because depends on engineer's skill, and pipe's natural frequency varies with its diameter, material amount flowing water. In order improve this situation necessary develop a sensor-based monitoring system leaks. Recently Fourier transform (FT) has been used, but insufficient when sound signal changes terms time,...

The theorems giving the conditions for discrete wavelet transforms (DWTs) to achieve perfect translation invariance (PTI) have already been proven, and based on these theorems, dual-tree complex DWT packet transform, achieving PTI, proposed. However, there is not so much flexibility in their density. In frequency domain, density fixed by octave filter banks, time each arrayed a coordinate, domain can be only designed dividing an band equally linear scale, its constrained division number of...

Echocardiography is widely used for the diagnosis of fetal cardiac arrhythmias. However, this method does not detect configurational changes in electrocardiogram (ECG) such as life-threatening QRS and prolongation QT interval. Fetal magnetocardiography (fMCG) electrocardiography (fECG) are valuable tools detection electrophysiological signals although both have certain limitations. Such techniques must deal with excess internal noise maternal respiratory movements, muscle contraction body...

Wavelet Shrinkage using DWT has been widely used in de-noising although a translation variance problem. In this study, we solve problem by the invariant DWT. For purpose, propose new complex wavelet, Real-Imaginary Spline (RI-Spline wavelet). We also Coherent Dual-Tree algorithm for RI-Spline wavelet and extend it to 2-Dimensional. Then apply de-noising. Experimental results show that our method, when applied ECG data, medical image textile surface inspection can obtain better than of...

The direction of a generated sound source can be estimated by humans using the time difference between both ears and characteristics sound. However, it is very difficult to create an estimation system computer two microphones. In this study, we propose novel method, in which feature recorded microphone signals used. As result, average correct rate 96% was obtained determining location 11 kinds object sounds

Adaboost algorithm is difficult to implement on embedded platform for real-time face detection by software due its high computation load and data throughput. This article presents a cell array architecture using parallel technology. Detection procedure can be greatly speeded up with multi- pipeline. Besides it makes use of the continuity image decrease accesses RAM. uses Electronic System Level (ESL) tools develop simulate cycle-accurate model architecture. The result shows that 200 MHz...

It is well known that the cellular neural network (CNN) very effective as an associative memory medium. And saturation (output) function plays important role in CNN, because it affects operation, stable equilibrium points and performance of CNN. However, to best our knowledge a systematic design procedure for output not available literature. In this paper, we present simple, yet, method two three-output functions. To demonstrate effectiveness functions, tested CNN on synthesized images....

In this paper, we propose a new complex-type wavelet which call the real imaginary (RI)-spline wavelet. We show that RI-spline can be used for continuous and discrete transform as mother confirm effectiveness of in cases by using model signal applying it to an unsteady analysis. The main results obtained are summarized follows: with both (CWT) (DWT) case DWT shift invariance has been improved. applied turbulence analysis removing pulse noise form music. experimental demonstrate advantages...

In this study, we examine the necessity for complex discrete wavelet transform (CDWT) from viewpoint that signal's position shifting corresponds to phase change and discuss effectiveness of improved shift invariability CDWT based on linear translation invariant system theory. The results are: 1) DWT is sensitive (the change) correspondingly coefficients sharply as signal shifts. On other hand, insensitive obtained are stable. 2) output by unique input having an arbitrary shift. That is, not...

In this paper, we propose a novel data augmentation method named Semantic Constraint Generative Adversarial Network (SCGAN) for person re-identification (Re-ID) in camera sensor networks. The proposed SCGAN can generate multiple style pedestrian images with high-level semantic information. To end, design two types of constraints, i.e., attention constraint and identity constraint. aims to restrict the significant areas map be consistent before after image transformation. focuses on keeping...