Chinese Journal of Liquid Crystals and Displays, Volume. 38, Issue 9, 1262(2023)
Laboratory flame image segmentation and recognition by fusing infrared and visible light
Fig. 1. Diagram of semantic perception of real-time infrared and visible image fusion division network structure
Fig. 2. Diagram of lightweight infrared and visible light image fusion network structure based on GRD
Fig. 8. Diagram of edge extraction module structure based on gradient transformation
Fig. 9. Comparison of fusion results with or without weights and IFTB module
Fig. 10. Comparison of segmentation results between the improved Deeplabv3+ network and the Deeplabv3+ infrastructure network
Table 1. Average estimation indicator of the weight module and IFTB module on the fusion result
View tableView in ArticleTable 1. Average estimation indicator of the weight module and IFTB module on the fusion result
评价指标 IFTB模块 权重模块 权重、IFTB模块 En 7.010 4 7.021 4 7.022 5 SF 9.637 1 8.784 6 9.596 1 VIFF 0.586 4 0.542 1 0.621 0 SCD 1.778 9 1.771 2 1.800 1 AG 3.952 1 3.763 5 3.956 3 0.701 3 0.736 4 0.736 6 0.479 9 0.469 5 0.508 0
Table 2. Comparison of flame smoke fusion image segmentation quantification results
View tableView in ArticleTable 2. Comparison of flame smoke fusion image segmentation quantification results
算法 MIoU/% FPS/（帧·s-1） Deeplabv3+基础网络 87.80 12.56 改进Deeplabv3+网络 91.27 11.96
Table 3. Flame segmentation and recognition results of improved Deeplabv3+ in different cases（MIoU）
View tableView in ArticleTable 3. Flame segmentation and recognition results of improved Deeplabv3+ in different cases（MIoU）
图像类别 烟雾遮挡/% 小火苗/% 可见光图像 80.49 79.34 融合图像 87.80 87.46
Get Citation
Copy Citation Text
Qi LI, Ran ZHANG. Laboratory flame image segmentation and recognition by fusing infrared and visible light[J]. Chinese Journal of Liquid Crystals and Displays, 2023, 38(9): 1262
Category: Research Articles
Received: Oct. 26, 2022
Accepted: --
Published Online: Sep. 19, 2023
The Author Email: Ran ZHANG (Zhangran0709@163.com)
