2025 Vol.8 Apr.N02

• Title: Research on multi-feature depth segmentation of 3D panoramic roaming images
• Name: Hongyan Lu1，Wanxin Liang1*，Mei Tang2，Cui Huang3
• Company: (1.Wuzhou Medical College，Wuzhou, Guangxi,543100,China (2.Wuzhou Vocational College,,Wuzhou, Guangxi,543000,China; (3.Guangxi University of Science and Technology，Liuzhou, Guangxi,545000,China
• Abstract:

  Traditional image segmentation techniques are often difficult to meet the high precision segmentation requirements of 3D panoramic roaming images. Therefore, the multi-feature depth segmentation technique of 3D panoramic roaming image is designed. The depth histogram is constructed, the gradient value of the image is calculated, and the Gaussian mixture model is constructed by using the foreground and background seed pixels marked by user interaction. The calculation of the foreground/background probability model is simplified by using Bayesian formula and equal probability hypothesis. By introducing the attention module, dimension splicing and weight allocation of the feature tensor are carried out to achieve multi-feature segmentation. The experimental results show that the design method has good segmentation results and can accurately determine the segmentation target location through multiple features.

  
  

• Keyword: Three-dimensional panorama; Roaming images; Multi-feature depth segmentation; Depth histogram; Attention module
• DOI: 10.12250/jpciams2025090408
• Citation form: Hongyan Lu,Wanxin Liang,Mei Tang,Cui Huang.Research on multi-feature depth segmentation of 3D panoramic roaming images[J]. Computer Informatization and Mechanical System,2025,Vol.8,pp.

[1]ZhangR , Chen C , Peng J .Multi-scale graph feature extraction network for panoramic image saliency detection[J].The visual computer, 2024, 40(2):953-970.

[2]Chen Y , Wang H , Gao J .A single defocused image depth recovery with superpixel segmentation[J].Pattern analysis and applications: PAA, 2023(3):26.

[3]Jin X , Zhu C , Li D .Panoramic image semantic segmentation using channel attention-based HarDNet and distorted boundary learning[J].Multimedia Systems, 2024, 30(6).

[4]Yu J , He Y , Zhang F ,et al.An Infrared Image Stitching Method for Wind Turbine Blade Using UAV Flight Data and U-Net[J].IEEE Sensors Journal, 2023, 23:8727-8736.

[5]Teng Z , Zhang J , Yang K ,et al.360BEV: Panoramic Semantic Mapping for Indoor Bird's-Eye View[J].ArXiv, 2023, abs/2303.11910.

[6]Dou Y , Yao F , Wang X ,et al.PanopticUAV: Panoptic Segmentation of UAV Images for Marine Environment Monitoring[J].CMES-Computer Modeling in Engineering & Sciences, 2024, 138(1).

[7]Li X , Wu T , Qi Z ,et al.SGAT4PASS: Spherical Geometry-Aware Transformer for PAnoramic Semantic Segmentation[J].ArXiv, 2023, abs/2306.03403.

[8]Priya J , Raja S K S , Sudha S .An intellectual caries segmentation and classification using modified optimization-assisted transformer denseUnet++and ViT-based multiscale residual denseNet with GRU[J].Signal, Image and Video Processing, 2024, 18(6-7):5213-5227.