Gaussian of Differences: A Simple and Efficient General Image Fusion Method

Entropy, Год журнала: 2023, Номер 25(8), С. 1215 - 1215

Опубликована: Авг. 15, 2023

The separate analysis of images obtained from a single source using different camera settings or spectral bands, whether one more than sensor, is quite difficult. To solve this problem, image containing all the distinctive pieces information in each generally created by combining images, process called fusion. In paper, simple and efficient, pixel-based fusion method proposed that relies on weighting edge associated with pixel proportional to distance their neighbors employing Gaussian filter. method, differences (GD), was evaluated multi-modal medical multi-sensor visible infrared multi-focus multi-exposure compared existing state-of-the-art methods utilizing objective quality metrics. parameters GD are further enhanced pattern search (PS) algorithm, resulting an adaptive optimization strategy. Extensive experiments illustrated ranked better average others terms metrics CPU time consumption.

Язык: Английский

---

Lightweight Tunnel Defect Detection Algorithm Based on Knowledge Distillation

Electronics, Год журнала: 2023, Номер 12(15), С. 3222 - 3222

Опубликована: Июль 26, 2023

One of the greatest engineering feats in history is construction tunnels, and management tunnel safety depends heavily on detection defects. However, real-time, portability, accuracy issues with present defect technique still exist. The study improves traditional technology based knowledge distillation algorithm, depth pooling residual structure designed teacher network to enhance ability extract target features. Next, MobileNetv3 lightweight built into student reduce number volume model parameters. then trained terms both features outputs using a multidimensional approach. By processing radar photos, dataset created. experimental findings demonstrate that approach greatly increases efficiency: parameters decreased by 81.4%, from 16.03 MB 2.98 MB, while improved 2.5%, 83.4% 85.9%.

Язык: Английский

---

MEEAFusion: Multi-Scale Edge Enhancement and Joint Attention Mechanism Based Infrared and Visible Image Fusion

Sensors, Год журнала: 2024, Номер 24(17), С. 5860 - 5860

Опубликована: Сен. 9, 2024

Infrared and visible image fusion can integrate rich edge details salient infrared targets, resulting in high-quality images suitable for advanced tasks. However, most available algorithms struggle to fully extract detailed features overlook the interaction of complementary across different modal during feature process. To address this gap, study presents a novel method based on multi-scale enhancement joint attention mechanism (MEEAFusion). Initially, convolution kernels varying scales were utilized obtain shallow with multiple receptive fields unique source image. Subsequently, gradient residual block (MGRB) was developed capture high-level semantic information low-level texture image, enhancing representation fine-grained features. Then, between defined, cross-transfer (CAFB) devised spatial channel refine critical supplemental information. This allowed network fused that both common information, thus realizing pre-fusion. Lastly, reconstructed Extensive experiments three benchmark datasets demonstrated MEEAFusion proposed research has considerable strengths terms details, significant distinct contours, it achieves superior performance.

Язык: Английский

---

Latest Advancements in Perception Algorithms for ADAS and AV Systems Using Infrared Images and Deep Learning

IntechOpen eBooks, Год журнала: 2023, Номер unknown

Опубликована: Дек. 7, 2023

Perception system plays an important role in Advanced driver assistance systems (ADAS) & Autonomous vehicles (AV) to understand the surrounding environment and further navigation. It is highly challenging achieve accurate perception of ego vehicle mimicking human vision. The available ADAS AV solutions could able perceive some extent using multiple sensors like Lidars, Radars Cameras. National Highway Traffic Safety Administration Crash reports shows that complete autonomy existing sensor suite. Particularly, extreme weather, low light night scenarios, there a need for additional sensors. Infrared camera seems be one potential address such corner cases. This chapter aimed discuss advantage adding infrared accurately. advancements deep learning approaches leverages enhance features. Also, limitations current sensors, technology, artificial intelligence research focus IR images are discussed detail. Literature by suite may lead way level 3 above autonomous driving precisely.

Язык: Английский

---

Recognition of Medical Images of Eye Diseases in a Deep Learning Perspective

2022 IEEE 10th Joint International Information Technology and Artificial Intelligence Conference (ITAIC), Год журнала: 2023, Номер unknown, С. 245 - 250

Опубликована: Дек. 8, 2023

By studying and learning various deep models, the most suitable model for identifying diseased eyes, normal highly myopic eyes in medical images is selected, effective identification prediction of eye are performed. Medical obtained through Alibaba Cloud's dataset analyzed image quality enhancement feature extraction. comparing test data between DenseNet121 finally used training. It can accurately identify different types make disease predictions based on images.

Язык: Английский

---

$$\hbox {KD}^{3}$$mt: knowledge distillation-driven dynamic mixer transformer for medical image fusion

The Visual Computer, Год журнала: 2024, Номер unknown

Опубликована: Окт. 1, 2024

Язык: Английский

---