Real-time Detection of Low-altitude Camouflaged Targets Based on Polarization Encoded Images

doi:10.12382/bgxb.2022.1289

Abstract

Abstract:

Polarization can improve the autonomous reconnaissance capability of unmanned aerial vehicle, but it is easily interfered by the variation of detection angle and target materials, which affects the robustness of polarization detection. In this paper, a real-time low-altitude camouflaged target detection algorithm of YOLO-Polarization based on polarized images is proposed. The coded image fused with multi-polarization direction information is used as input, the 3D convolution module is applied to extract the connection features from the different polarization direction images, and a feature enhancement module (FEM) is introduced to further enhance the multi-level features. In addition, the cross-level feature aggregation network is adopted to make full use of the feature information of different scales to complete the effective aggregation of features, and finally combined with multi-channel feature information output detection results. A dataset consisting of polarized images of low-altitude camouflaged targets (PICO) which include 10 types of targets is constructed. The experimental results based on PICO dataset show that the proposed method can effectively detect the camouflaged targets, with mAP_0.5:0.95 up to 52.0% and mAP_0.5 up to 91.5%. The detection rate achieves 55.0 frames/s, which meets the requirement of real-time detection.

Key words: unmanned aerial vehicle, camouflaged target detection, deep learning, polarization imaging, feature enhancement, feature aggregation

CLC Number:

SHEN Ying, LIU Xiancai, WANG Shu, HUANG Feng. Real-time Detection of Low-altitude Camouflaged Targets Based on Polarization Encoded Images[J]. Acta Armamentarii, 2024, 45(5): 1374-1383.

Figures/Tables 9

Fig.1 Overall architecture of YOLO-P network

Fig.2 Structure of feature enhancement module

Fig.3 Cross-level feature aggregation structure

Fig.4 Polarization encoded image of camouflaged target

Table 1 Comparison of different polarized images of the same target under different imaging conditions

Table 2 Comparison of input detection results of different polarized images

数据输入	mAP_0.5/%	mAP_0.5:0.95/%
S₀	77.4	39.5
DoLP	43.4	20.6
I_S	45.0	21.0
I₀、I₄₅、I₉₀	85.5	46.0

Table 3 Comparison of test results of different models

网络结构	mAP_0.5/%	mAP_0.5:0.95/%	FPS/(帧·s^-1)
Faster R-CNN	63.0	30.1	7.8
YOLOv4-S	75.0	42.0	41.0
YOLOv5-S	77.7	38.5	73.5
YOLOX-S	85.5	46.0	63.0
YOLOv7	87.8	48.0	76.0
YOLO-P	91.5	52.0	55.0

Table 4 Partial test results of different models

Table 5 Results of ablation experiment

序号	预训练权重	3D卷积模块	FEM	特征聚合网络	mAP_0.5/ %	参数量/ M
1					80.5	8.9
2	√				85.5	8.9
3	√	√			86.2	8.9
4	√		√		88.5	9.9
5	√			√	87.0	8.8
6	√	√	√	√	91.5	9.8

References 27

[1]	HUANG Y Q, DING W R, LI H G. Haze removal for UAV reconnaissance images using layered scattering model[J]. Chinese Journal of Aeronautics, 2016, 29(2): 502-511.
[2]	GAO S, WU J Z, AI J L. Multi-UAV reconnaissance task allocation for heterogeneous targets using grouping ant colony optimization algorithm[J]. Soft Computing, 2021, 25(10): 7155-7167.
[3]	YANG X, XU W D, JIA Q, et al. MF-CFI: a fused evaluation index for camouflage patterns based on human visual perception[J]. Defence Technology, 2021, 17(5):1602-1608. doi: 10.1016/j.dt.2020.08.007
[4]	BI H B, ZHANG C, WANG K, et al. Rethinking camouflaged object detection: models and datasets[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 32(9):5708-5724.
[5]	MONDAL A. Camouflaged object detection and tracking:a survey[J]. International Journal of Image & Graphics, 2020, 20(4):2050028.
[6]	XUE F, CUI G Y, HONG R C, et al. Camouflage texture evaluation using a saliency map[J]. Multimedia Systems, 2015, 21(2): 165-175.
[7]	ZHANG X, ZHU C, WANG S, et al. A Bayesian approach to camouflaged moving object detection[J]. IEEE Transactions on Circuits & Systems for Video Technology, 2017, 27(9): 2001-2013.
[8]	JR H, CUTHILL I C, BADDELEY R, et al. Camouflage, detection and identification of moving targets[J]. Proceedings of the Royal Society B-Biological Sciences, 2013, 280(1758): 20130064.
[9]	FAN D P, JI G P, SUN G, et al. Camouflaged object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York,NY, US:IEEE,2020: 2777-2787.
[10]	WANG K, BI H B, ZHANG Y, et al. D²C-net: a dual-branch, dual-guidance and cross-refine network for camouflaged object detection[J]. IEEE Transactions on Industrial Electronics, 2021, 69(5): 5364-5374.
[11]	ZHOU T, ZHOU Y, GONG C, et al. Feature aggregation and propagation network for camouflaged object detection[J]. IEEE Transactions on Image Processing, 2022, 31:7036-7047. doi: 10.1109/TIP.2022.3217695 pmid: 36331642
[12]	MEI H Y, JI G P, WEI Z Q, et al. Camouflaged object segmentation with distraction mining[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville, TN, US: IEEE, 2021: 8772-8781.
[13]	ZHANG J C, SHAO J B, CHEN J L, et al. Polarization image fusion with self-learned fusion strategy[J]. Pattern Recognition, 2021, 118: 108045.
[14]	TANG F Z, GUI L Q, LIU J B, et al. Metal target detection method using passive millimeter-wave polarimetric imagery[J]. Optics Express, 2020, 28(9):13336-13351. doi: 10.1364/OE.390385 pmid: 32403811
[15]	BLIN R, AINOUZ S, CANU S, et al. The PolarLITIS dataset: road scenes under fog[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(8):10753-10762.
[16]	WANG H F, SHAN Y H, HAO T, et al. Vehicle-road environment perception under low-visibility condition based on polarization features via deep learning[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(10): 17873-17886.
[17]	LIANG J A, WANG X, FANG Y J, et al. Water surface-clutter suppression method based on infrared polarization information[J]. Applied Optics, 2018, 57(16):4649-4658.
[18]	ZHANG Y, ZHANG Y, ZHAO H J, et al. Improved atmospheric effects elimination method for pBRDF models of painted surfaces[J]. Optics Express, 2017, 25(14): 16458-16475. doi: 10.1364/OE.25.016458 pmid: 28789150
[19]	SHEN Y, LIN W F, WANG Z F, et al. Rapid detection of camouflaged artificial target based on polarization imaging and deep learning[J]. IEEE Photonics Journal, 2021, 13(4):1-9.
[20]	REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149. doi: 10.1109/TPAMI.2016.2577031 pmid: 27295650
[21]	GE Z, LIU S T, WANG F, et al. Yolox: exceeding yolo series in 2021:arXiv:2107.08430[R]. Ithaca,NY, US: Cornell Universit, 2021:2107.08430.
[22]	BOCHKOVSKIY A, WANG C, LIAO H M. YOLOv4:optimal speed and accuracy of object detection: arXiv:2004.10934[R]. Ithaca,NY, US: Cornell Universit, 2020:2004.10934.
[23]	WOO S Y, PARK J C, LEE J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2018:3-19.
[24]	SUN R, SUN X B, CHEN F, et al. Polarimetric imaging detection using a convolutional neural network with three-dimensional and two-dimensional convolutional layers[J]. Applied Optics, 2020, 59(1): 151-155.
[25]	谭湘粤, 胡晓, 杨佳信, 等. 基于递进式特征增强聚合的伪装目标检测[J]. 计算机应用, 2022, 42(7):2192-2200. doi: 10.11772/j.issn.1001-9081.2021060900
	TAN X Y, HU X, YANG J X, et al. Camouflaged object detection based on progressive feature enhancement aggregation[J]. Journal of Computer Applications, 2022, 42(7): 2192-2200. (in Chinese) doi: 10.11772/j.issn.1001-9081.2021060900
[26]	TAN M X, PANG R M, LE Q V. Efficientdet: scalable and efficient object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle, WA, US:IEEE, 2020:10781-10790.
[27]	WANG C Y, BOCHKOVSKIY A, LIAO H Y M. Yolov7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors:arXiv:2207.02696[R]. Ithaca,NY, US: Cornell Universit, 2022:2207.02696. 2022.