IS&T | Library

Digital Library: JIST

Published Online: July 2026

Efficient and Robust Semantic Segmentation Method based on Multi-Scale Feature Fusion and Temporal Alignment

5 1

temporal alignment
multi-modal fusion
semantic segmentation
autonomous driving

Shuyi Tan, Chao Huang, Yi Zhang, Yang Wang

Pages 1 - 11, July 2026, © Society for Imaging Science and Technology 2026

DOI

10.2352/J.ImagingSci.Technol.2026.70.4.040501

Volume 70

Issue 4

Abstract

Owing to its ability to enable precise perception of dynamic and complex environments, point cloud semantic segmentation has become a critical task for autonomously driven vehicles in recent years. However, in complex, dynamic scenes, cumulative errors pose significant challenges for existing semantic segmentation methods, limiting their accuracy and efficiency, particularly in safety-critical applications. To address these issues, this paper introduces a novel framework that balances accuracy and computational efficiency by leveraging temporal alignment. The framework effectively captures inter-frame correlations, enhances local detail information, reduces error accumulation, and maintains detailed scene features. Furthermore, by integrating LiDAR and camera data through multi-modal fusion, the framework provides complementary perspectives, significantly improving segmentation performance and robustness in dynamic environments. This method achieves competitive performance on the benchmark SemanticKITTI and nuScenes datasets, demonstrating its capability to detect occluded objects and ensure reliable perception in safety-critical scenarios. The proposed framework offers a promising solution for enhancing the robustness and reliability of autonomous driving systems in complex environments.

Digital Library: JIST

Published Online: July 2026

Spatiotemporal Trend–Event Decoupling Mamba Graph Network for Traffic Flow Forecasting

11 1

traffic flow forecasting
spatiotemporal modeling
trend–event decoupling
dynamic graph convolution
selective state space model
causal convolution

Rui An, Jianxin Cao, Tengfei Zhao

Pages 1 - 15, July 2026, © Society for Imaging Science and Technology 2026

DOI

10.2352/J.ImagingSci.Technol.2026.70.4.040502

Volume 70

Issue 4

Abstract

Traffic flow forecasting faces substantial challenges arising from intertwined heterogeneous temporal dynamics and complex spatial dependencies. A key difficulty lies in balancing long-term stable trends with short-term abrupt disturbances. To address these challenges, a framework named the Spatiotemporal Trend–Event Decoupling Mamba Graph Network (STEDMGN) is proposed. First, a temporal signal separation layer is constructed using a multi-scale decomposition and reconstruction mechanism to divide the raw sequence into trend and event components. This design enables dynamic pattern decoupling across multiple scales. Subsequently, a dual-frequency spatiotemporal encoder is introduced. The trend branch integrates multi-head attention with a Mamba-based state space layer to capture cross-period long-term dependencies, whereas the event branch employs causal convolution to model short-term abrupt disturbances. In the spatial dimension, trend-oriented and event-oriented graph convolutional networks are incorporated. These networks combine static priors, adaptive adjacency, and feature-driven dynamic graph structures to enhance the representation of both stable topologies and time-varying propagation. Finally, a fusion-gated decoder employs gated units and a query-driven fusion strategy to integrate the two feature types. A subsequent regression layer then generates multi-step forecasts. Extensive experiments on four public PeMS datasets demonstrate that the STEDMGN substantially outperforms state-of-the-art methods. The results provide an accurate and scalable solution for large-scale urban traffic flow forecasting.

Digital Library: JIST

Published Online: July 2026

MonoHybrid: Self-Supervised Monocular Depth Estimation with Hybrid Network

19 1

monocular depth estimation
hybrid network
self-supervision

Wei Jiang, Bingfei Nan, Guofa Wang

Pages 1 - 12, July 2026, © Society for Imaging Science and Technology 2026

DOI

10.2352/J.ImagingSci.Technol.2026.70.4.040503

Volume 70

Issue 4

Abstract

Monocular depth estimation (MDE) is a widely used technique in autonomous driving and 3D reconstruction. However, inconsistent and fragmented depth outputs can significantly undermine the reliability of MDE applications in practice. To address this issue, the authors introduce MonoHybrid, a novel self-supervised hybrid network that effectively integrates Transformer and dilated convolutional architectures. This design enables the extraction of both global and local features, enhancing the receptive field and ensuring robust and continuous depth estimation. Additionally, the authors present a new Feature Fusion Module that fuses convolutional and Transformer features, resulting in improved depth estimation performance. Through comprehensive experiments, the proposed network demonstrates notable accuracy and generalization compared to other advanced methods in the field.

Digital Library: JIST

Published Online: July 2026

Human Dynamic Behavior Recognition Method based on Joint Point Extraction and Deep Learning Algorithm

8 1

recognition of human dynamic behavior
3D skeletal joint point cloud; deep learning; PointNet++ method

Lei Ma, Yunwei Li

Pages 1 - 14, July 2026, © Society for Imaging Science and Technology 2026

DOI

10.2352/J.ImagingSci.Technol.2026.70.4.040504

Volume 70

Issue 4

Abstract

This study proposes a human dynamic behavior recognition method based on joint point extraction and deep learning algorithms. Human skeletal information is collected using a Kinect camera to obtain three-dimensional (3D) joint coordinates, completing the extraction of skeletal joint data. Based on the characteristics of bones in 3D space, processing is performed using 3D skeletal joint point cloud data. An improved PointNet++ method is employed to process the point cloud data: a dual-scale feature extraction strategy is adopted to enhance the network’s multiscale feature capture capability, and the farthest point sampling algorithm is optimized to better preserve behavioral details. Finally, a graph convolutional network approach incorporating a channel attention mechanism and graph topology optimization is used to achieve human dynamic behavior recognition based on 3D skeletons. Experimental results show that the method achieves a maximum F₁ score of 0.98, a misrecognition rate below 0.4%, and a coefficient of variation of 0, demonstrating high recognition accuracy. However, the performance of this research method may be limited in complex scenes with severe occlusion or non-standard perspectives. Future work will focus on exploring multimodal data fusion and real-time optimization to further enhance its robustness and practicality in open environments.

Digital Library: JIST

Published Online: July 2026

Soft Prototype Contrastive Learning and Instance Discrimination for Unsupervised Visible–Infrared Person Re-Identification

11 1

person re-identification
unsupervised learning
visible–infrared
contrastive learning

Mengke Li, Jianwei Yang

Pages 1 - 10, July 2026, © Society for Imaging Science and Technology 2026

DOI

10.2352/J.ImagingSci.Technol.2026.70.4.040505

Volume 70

Issue 4

10.2352/J.ImagingSci.Technol.2026.70.3.030101

Abstract

Unsupervised visible–infrared person re-identification (USVI-ReID) is a very important and challenging task in machine vision. The key challenge of USVI-ReID is to effectively mine weak class-wise supervision and establish cross-modal correspondences without using any manual annotations. In this paper, the authors propose a soft prototype contrastive learning and instance discrimination method for USVI-ReID. Specifically, soft prototype contrastive learning selects the nearest neighbors with high similarity to the soft prototypes to mine accurate information and guide the model to learn more discriminative features. On this basis, a soft weighting strategy is used to quantitatively measure the relevance of the selected soft prototypes relative to the current centroid prototype, thus further eliminating the interference of the wrong prototype in the model training. To overcome the problems of image noise and complex backgrounds in visible and infrared images, instance discriminative learning is first integrated into USVI-ReID to explore the potential similarity relationship between instances from the bottom up and learn discriminative representations. Finally, the authors propose a progressive training strategy, which enables the model to learn the similarities between instances in the early stage of training and gradually shift its attention to more discriminative categories in the later stage. Extensive experiments are conducted on two public datasets, and quantitative results prove the effectiveness of the proposed method.

Digital Library: JIST

Published Online: July 2026

Editorial

119 1

From the Editor

Chunghui Kuo

DOI

Volume 70

Issue 3

Silver-Plated Polymer Horn Antenna for K-Band Applications Using Precision 3D Printing

Digital Library: JIST

Published Online: May 2026

Special issue CAPT 2025

165 3

corrugated horn antenna
K-band
high precision
additive manufacturing
chemical silver plating
3D printing

Yukang Liu, Yingjie Xu, Qunxing Bai, Wan Zhang, Chang Xu

DOI

10.2352/J.ImagingSci.Technol.2026.70.3.030401

Volume 70

Issue 3

Optimized Construction of Ink Base Databases for Accurate Spot Color Matching Using Single-Constant Kubelka–Munk Model

Abstract

This research develops a novel manufacturing approach for millimeter-wave feedhorns, utilizing additive manufacturing combined with electroless metallization. A corrugated horn antenna operating across the K-band spectrum was engineered and produced using polymer-based 3D printing, followed by internal surface silver deposition. This methodology achieved complex internal geometry consolidation in a single process, yielding a structure with merely 17% the mass of comparable steel counterparts. Electrical characterization demonstrated reflection coefficients predominantly exceeding −20 dB magnitude across 18.0–27.0 GHz alongside the attained gain values surpassing 14 dB within 18.0–24.0 GHz. The measured far-field radiation characteristics showed excellent correlation with computational electromagnetic models. The demonstrated technique presents transformative potential for the mass-efficient production of high-frequency components in next-generation small satellite constellations and compact radar platforms.

Digital Library: JIST

Published Online: May 2026

Special issue CAPT 2025

114 1

spot color matching
Kubelka–Munk theory
ink base database
spectral similarity
K/S linearity

Bo Liu, Yujiao Du, Miaoxin Li, Qifan Jia, Shiwei Liu, Chun-ao Wei, Junfeng Li

DOI

10.2352/J.ImagingSci.Technol.2026.70.3.030402

Volume 70

Issue 3

Research on Tension Control Strategy for Unwinding System of Lithium Battery Separator Slitting Machine based on Immune Genetic Algorithm-Optimized Active Disturbance Rejection Control

Abstract

Accurate spot color matching is critical to printing applications, yet constructing an efficient ink base database remains a challenge due to the labor-intensive preparation of ink ladder samples. This study proposes a two-step optimization method to enhance the efficiency and accuracy of spot color prediction using the single-constant Kubelka–Munk (KM) model. The first step employs spectral similarity screening via the Goodness-of-Fit Coefficient to select samples with consistent spectral behavior. The second step optimizes for K/S linearity, identifying concentrations (35% and 40%) that best align with the KM model’s linearity assumption. Five target spot colors, created by mixing yellow, red, and blue base inks, were used to evaluate the method. The K/S values derived from three sample sets—all ladder samples, one-step optimized samples, and two-step optimized samples—were used to predict spectral reflectance and CIE Lab values, with color differences (ΔE) calculated against measured values. The two-step optimized samples achieved the lowest average ΔE value of 3.08 compared to 7.38 for all samples and 4.59 for one-step optimized samples, demonstrating superior accuracy. By reducing the required samples from 19 to 2 per ink, the method significantly enhances efficiency without compromising precision. These findings highlight the importance of spectral consistency and K/S linearity for reliable color matching and offer a practical solution for industrial applications such as packaging and branding.

Digital Library: JIST

Published Online: May 2026

Special issue CAPT 2025

108 2

lithium battery separator slitting machine
unwinding system
immune genetic algorithm-optimized active disturbance rejection control
tension control

Jian Li, Jun Yuan, Kui He, Haitao Hao, XuChen Zhang, Pengbo He

DOI

10.2352/J.ImagingSci.Technol.2026.70.3.030403

Volume 70

Issue 3