Recently, vein recognition has been paid more considerable attention in biometric fields. In the process of image acquisition, due to influence external factors such as illumination change, texture information images with same identity may which enhances difference intra-class and extremely degrades performance systems. To address this problem, we proposed a Disentangled Representation Enhancement Network for called DRE-Net. First, robust shape masks are obtained by designed segmentation...

RGB and thermal image fusion have great potential to exhibit improved semantic segmentation in low-illumination conditions. Existing methods typically employ a two-branch encoder framework for multimodal feature extraction design complicated strategies achieve segmentation. However, these require massive parameter updates computational effort during the fusion. To address this issue, we propose novel network (EFNet) based on an early strategy simple but effective clustering training...

Pre-trained DCNN trained on a large-scale image database can be used as universal feature representation for classification, which has achieved significant progress in some recognition tasks. Compared with other tasks, directly utilizing single convolutional vein task cannot achieve the impressive result due to sparse distribution of information. Therefore, obtain more representative and discriminative recognition, novel multi-layer features' concatenation semantic selector is proposed this...

Abstract Background Both patient-specific dose recalculation and γ passing rate analysis are important for the quality assurance (QA) of intensity modulated radiotherapy (IMRT) plans. The aim this study was to analyse correlation between rates volumes air cavities ( V ) bony structures bone in target volume head neck cancer. Methods Twenty nasopharyngeal carcinoma twenty nasal natural killer T-cell lymphoma patients were enrolled study. Nine-field sliding window IMRT plans produced...

Stereotactic radiotherapy (SRT) has been widely used for the treatment of brain metastases and early stage non-small-cell lung cancer (NSCLC). Excellent SRT plans are characterized by steep dose fall-off, making it critical to accurately comprehensively predict evaluate fall-off.A novel fall-off index was proposed ensure high-quality planning.The gradient (NGI) had two different modes: NGIx V three-dimensions r one-dimension. were defined as ratios decreased percentage (x%) corresponding...

In recent years, various applications in computer vision have achieved substantial progress based on deep learning, which has been widely used for image fusion and shown to achieve adequate performance. However, suffering from limited ability modeling the spatial correspondence of different source images, it still remains a great challenge existing unsupervised models extract appropriate feature achieves adaptive balanced fusion. this paper, we propose novel cross-attention-guided network,...

Lung 4D computed tomography (4D-CT) plays an important role in high-precision radiotherapy because it characterizes respiratory motion, which is crucial for accurate target definition. However, the manual segmentation of a lung tumor heavy workload doctors large number 4D-CT data slices. Meanwhile, still notoriously challenging problem computer-aided diagnosis. In this paper, we propose new method based on improved graph cut algorithm with context information constraint to find convenient...

3D deblurring reconstruction techniques have recently seen significant advancements with the development of Neural Radiance Fields (NeRF) and Gaussian Splatting (3DGS). Although these can recover relatively clear reconstructions from blurry image inputs, they still face limitations in handling severe blurring complex camera motion. To address issues, we propose Event-assisted Deblur Reconstruction (EaDeblur-GS), which integrates event data to enhance robustness 3DGS against motion blur. By...

Rain streaks vary in size, direction, and density, resulting serious blurring image quality degradation, which often directly affect the downstream visual tasks. At present, many end-to-end removal networks have achieved good results, but details are lost during processing. Therefore, we propose a novel detail-recovery network to solve this problem. Unlike existing works, regard rain detail restoration as two different tasks simultaneously. Specifically, use encoder-decoder extract detailed...