带式输送机作为煤炭运输系统的核心装备，其智能化节能调速技术成为实现行业降本增效的关键突破点。针对煤矿带式输送机节能调速控制需求，传统煤流量检测方法在中小型煤矿企业应用中面临成本高、实时性不足等问题。本文聚焦基于视觉的煤流动态占比检测，针对复杂工况导致煤流感知精度受限的难题，设计了低照度增强算法与实时语义分割算法级联的煤流动态占比检测算法。本文主要工作如下：

（1）针对带式输送机监控视频图像照度低、照度不均匀、光晕伪影等问题，本文提出一种高效的改进Zero-DCE低照度图像增强算法。首先，采用GhostNetV3模块轻量化曲线参数估计网络，大幅度减小模型的参数量与计算量。其次，将空间一致性损失扩展至八方向梯度约束，增强局部光照连续性。此外，构建多尺度结构感知损失强化边缘细节保留。最后，引入自适应颜色分布损失与频域小波损失，在Lab色彩空间约束色度均衡性，并通过哈尔小波分解同步优化高频纹理与低频结构。实验表明，提出的改进Zero-DCE算法在煤流分割数据集上效果表现出色，且在客观和主观两个评价方面均优于其他低照度图像增强算法。

（2）针对传统煤流量检测方法在动态敏感性与环境适应性上的不足，本文提出一种基于计算机视觉的非接触式煤流动态占比检测算法。首先，通过构建煤流占比的二维空间表征模型，揭示深度学习语义分割在特征表征与场景泛化方面的技术优势。其次，针对工业场景中低照度干扰与实时性需求，设计双分支实时语义分割网络DENet，通过多尺度通道注意力模块提取多尺度语义特征，结合细节增强模块强化细节表征能力，并设计无参数注意力引导的增强模块实现跨层级语义-细节特征融合，设计特征融合模块，自适应地融合不同层级的特征图。最后，经实验表明，DENet在煤流分割数据集上实现96.23%的mIoU与87.1 FPS实时性能，为带式输送机能耗优化提供了高精度、低成本的视觉解决方案。

（3）针对带式输送机煤流动态占比检测需求，本文提出“增强-分割-计算”级联检测框架，集成改进Zero-DCE低照度增强算法与DENet语义分割网络。通过改进Zero-DCE算法实现低照度图像增强，结合DENet网络完成煤流和皮带精准分割，构建端到端检测框架。实验表明，图像增强后分割精度提升1.48%，煤流占比误差由2.21%降至1.34%；交叉验证显示改进Zero-DCE与DENet组合在斜井过曝、井下弱光等极端场景下表现最优。通过TensorRT优化实现嵌入式部署，在Jetson Nano平台达到31.2 FPS实时性能，验证工业场景适用性。

As the core equipment of coal transportation system, the intelligent energy-saving speed control technology of belt conveyor has become a key breakthrough point to realize the industry's cost reduction and efficiency. For the demand of energy-saving speed control of coal mine belt conveyor, traditional coal flow detection method faces high cost and lack of real-time in the application of small and medium-sized coal mining enterprises. In this paper, focusing on vision-based coal flow dynamic proportion detection, aiming at the problem of limited coal flow sensing accuracy due to the complex working conditions, designed dynamic proportion of coal flow detection algorithm with low illumination enhancement algorithm cascaded with real-time semantic segmentation algorithm. The main work of this paper is as follows:

(1) Aiming at the problems of low illumination, uneven illumination and halo artifacts in the video images of belt conveyor monitoring, this paper proposes an efficient and improved Zero-DCE low illumination image enhancement algorithm. First, the GhostNetV3 module is used to lighten the curve parameter estimation network, which greatly reduces the number of parameters and computation of the model. Second, the spatial consistency loss is extended to eight-direction gradient constraints to enhance the local illumination continuity. In addition, the multi-scale structure-aware loss is constructed to enhance the edge detail reservation. Finally, the adaptive color distribution loss and frequency domain wavelet loss are introduced to constrain the chromatic balance in Lab color space and optimize the high-frequency texture and low-frequency structure synchronously through the Haar wavelet decomposition. Experiments show that the proposed improved Zero-DCE algorithm performs well on the coal flow segmentation dataset and outperforms other low-light image enhancement algorithms in both objective and subjective evaluations.

(2) Aiming at the shortcomings of traditional coal flow detection methods in terms of dynamic sensitivity and environmental adaptability, this paper proposes a non-contact coal flow dynamic proportion detection algorithm based on computer vision. First, by constructing a two-dimensional spatial representation model of coal flow proportion, the technical advantages of deep learning semantic segmentation in feature representation and scene generalization are revealed. Second, for the low illumination interference and real-time demand in industrial scenes, design a two-branch real-time semantic segmentation network DENet, extract multi-scale semantic features through multi-scale channel attention module, strengthen detail representation ability by combining with detail enhancement module, design parameter-free attention-guided enhancement module to realize cross-level semantic-detail feature fusion, and design a feature fusion module to adaptively fuse different layers of the of feature maps. Finally, it is experimentally shown that DENet achieves 96.23% mIoU with 87.1 FPS real-time performance on the coal flow segmentation dataset, which provides a high-precision and low-cost vision solution for optimizing the energy consumption of belt conveyor.

(3) In order to meet the demand of dynamic proportion detection of coal flow in belt conveyor, this paper proposes a cascade detection framework of “enhancement-segmentation-computation”, which integrates the improved Zero-DCE low illumination enhancement algorithm and DENet semantic segmentation network. The improved Zero-DCE algorithm realizes low illumination image enhancement, combines with DENet network to complete accurate segmentation of coal flow and belt, and builds an end-to-end detection framework. Experiments show that the segmentation accuracy is improved by 1.48% after image enhancement, and the error of coal flow percentage is reduced from 2.21% to 1.34%; cross-validation shows that the combination of improved Zero-DCE and DENet performs optimally in extreme scenarios such as overexposure of inclined shafts and low light downhole. Embedded deployment is achieved through TensorRT optimization, and 31.2 FPS real-time performance is achieved on Jetson Nano platform to verify the applicability of industrial scenarios.

[1] 王海军,王洪磊.带式输送机智能化关键技术现状与展望[J].煤炭科学技术,2023,50(12):225-239.

[2] 周坪,马国庆,周公博,等.智能化带式输送机健康监测技术研究综述[J].仪器仪表学报,2024,44(12):1-21.

[3] 王忠鑫,田凤亮,孙鑫,等.露天煤矿智能化能力成熟度与演进逻辑研究[J].工矿自动化,2025,51(1):119-125.

[4] Suchorab N. Specific energy consumption–the comparison of belt conveyors[J]. Mining Science,2019,26:263-274.

[5] Kawalec W, Suchorab N, Konieczna-Fuławka M, et al. Specific energy consumption of a belt conveyor system in a continuous surface mine[J]. Energies,2020,13(19):5214.

[6] Soofastaei A, Karimpour E, Knights P, et al. Energy-efficient loading and hauling operations[J]. Energy efficiency in the minerals industry: best practices and research directions,2018:121-146.

[7] 王传奇.带式输送机驱动系统技术现状及发展趋势[J].现代制造技术与装备,2022(001):058.

[8] 种磊,李智源.煤矿智能带式输送机运输系统关键技术综述[J].煤炭工程,2022,54(S01):32-36.

[9] 黄向东,邓莉洁.一种基于Retinex的矿井非均匀照度图像增强算法[J].科学技术与工程,2014,14(05):141-144+170.

[10] 杨勇,岳建华,李玉良,等.一种矿井动态图像增强方法[J].工矿自动化,2015,41(11):48-52.

[11] 程德强,郑珍,姜海龙.一种煤矿井下图像增强算法[J].工矿自动化,2015,41(12):31-34.

[12] Chai Y, Gao R, Deng L J. Study of image enhancement algorithms in coal mine[C]//2016 8th International Conference on Intelligent Human-Machine Systems and Cybernetics(IHMSC),2016:383-386.

[13] 谢海波.提升小波变换域矿井光照不均匀图像双直方图均衡化增强[J].金属矿山,2016,No.479(05):153-157.

[14] 刘晓阳,乔通,乔智.基于双边滤波和Retinex算法的矿井图像增强方法[J].工矿自动化,2017,43(02):49-54.

[15] 王星,白尚旺,潘理虎,等.一种矿井图像增强算法[J].工矿自动化,2017,43(03):48-52.

[16] 付燕,李瑶,严斌斌.一种煤矿井下视频图像增强算法[J].工矿自动化,2018,44(07):80-83.

[17] Wu D, Zhang S. Research on image enhancement algorithm of coal mine dust[C]//2018 International Conference on Sensor Networks and Signal Processing(SNSP),2018:261-265.

[18] 王树奇,刘贝,邹斐.一种新的矿井监控视频增强目标检测算法[J].西安科技大学学报,2019,39(02):347-353.

[19] 田子建,王满利,吴君,等.基于双域分解的矿井下图像增强算法[J].光子学报,2019,48(05):107-119.

[20] 朱礼义,李巧月,李国超,等.基于HSI空间融合的矿井图像增强算法[J].计算机工程与设计,2019,40(10):2926-2930+3008.

[21] 张静.基于视频的煤矿井下人形目标检测[D].西安:西安科技大学,2019.

[22] 赵谦,钱渠,任志奇.BEMD分解的矿下图像增强算法[J].西安科技大学学报,2020,40(03):484-491.

[23] 王满利,田子建.基于非下采样轮廓波变换的矿井图像增强算法[J].煤炭学报,2020,45(09):3351-3362.

[24] Guo C, Li C, Guo J, et al. Zero-reference deep curve estimation for low-light image enhancement[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2020:1780-1789.

[25] 王诚聪,刘亚静.矿井复杂环境视频监控图像增强算法研究[J].煤炭工程,2021,53(04):147-151.

[26] 王媛彬,韦思雄,段誉,等.基于自适应双通道先验的煤矿井下图像去雾算法[J].工矿自动化,2022,48(5):7.

[27] 王满利,张航,李佳悦,等.基于深度神经网络的煤矿井下低光照图像增强算法[J].煤炭科学技术,2023,51(9):231-241.

[28] Zheng Z, Chuah M C. Latent Disentanglement for Low Light Image Enhancement[C]//2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE,2024:2728-2733.

[29] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2015:3431-3440.

[30] Badrinarayanan V, Kendall A, Cipolla R. Segnet: A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE transactions on pattern analysis and machine intelligence, 2017,39(12):2481-2495.

[31] Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C]//Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9,2015, proceedings, part III 18. Springer International Publishing,2015:234-241.

[32] Chen L C. Rethinking atrous convolution for semantic image segmentation[J]. arXiv preprint arXiv:1706.05587,2017.

[33] Chen L C, Zhu Y, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Proceedings of the European conference on computer vision (ECCV).2018:801-818.

[34] Zhao H, Shi J, Qi X, et al. Pyramid scene parsing network[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2017:2881-2890.

[35] Wang J, Sun K, Cheng T, et al. Deep high-resolution representation learning for visual recognition[J]. IEEE transactions on pattern analysis and machine intelligence,2020,43(10):3349-3364.

[36] Orsic M, Kreso I, Bevandic P, et al. In defense of pre-trained imagenet architectures for real-time semantic segmentation of road-driving images[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2019:12607-12616.

[37] Deng J, Dong W, Socher R, et al. Imagenet: A large-scale hierarchical image database[C]//2009 IEEE conference on computer vision and pattern recognition. IEEE,2009:248-255.

[38] Fan M, Lai S, Huang J, et al. Rethinking bisenet for real-time semantic segmentation[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2021:9716-9725.

[39] Yu C, Wang J, Peng C, et al. Bisenet: Bilateral segmentation network for real-time semantic segmentation[C]//Proceedings of the European conference on computer vision (ECCV).2018:325-341.

[40] Poudel R P K, Liwicki S, Cipolla R. Fast-scnn: Fast semantic segmentation network[J]. arXiv preprint arXiv:1902.04502,2019.

[41] Gamal M, Siam M, Abdel-Razek M. Shuffleseg: Real-time semantic segmentation network[J]. arXiv preprint arXiv:1803.03816,2018.

[42] Zhang X, Zhou X, Lin M, et al. Shufflenet: An extremely efficient convolutional neural network for mobile devices[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2018:6848-6856.

[43] Li X, You A, Zhu Z, et al. Semantic flow for fast and accurate scene parsing[C]//Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part I 16. Springer International Publishing,2020:775-793.

[44] Yu C, Gao C, Wang J, et al. Bisenet v2: Bilateral network with guided aggregation for real-time semantic segmentation[J]. International journal of computer vision,2021,129:3051-3068.

[45] Poudel R P K, Bonde U, Liwicki S, et al. Contextnet: Exploring context and detail for semantic segmentation in real-time[J]. arXiv preprint arXiv:1805.04554,2018.

[46] Pan H, Hong Y, Sun W, et al. Deep dual-resolution networks for real-time and accurate semantic segmentation of traffic scenes[J]. IEEE Transactions on Intelligent Transportation Systems,2022,24(3):3448-3460.

[47] Liu Z, Hao Z, Han K, et al. GhostNetV3: Exploring the Training Strategies for Compact Models. arXiv 2024[J]. arXiv preprint arXiv:2404.11202.

[48] Mertens T, Kautz J, Van Reeth F. Exposure fusion[C]//15th Pacific Conference on Computer Graphics and Applications (PG'07). IEEE,2007:382-390.

[49] Mertens T, Kautz J, Van Reeth F. Exposure fusion: A simple and practical alternative to high dynamic range photography[C]//Computer graphics forum. Oxford, UK: Blackwell Publishing Ltd,2009,28(1):161-171.

[50] Li C, Guo C, Loy C C. Learning to enhance low-light image via zero-reference deep curve estimation[J]. IEEE transactions on pattern analysis and machine intelligence,2021,44(8):4225-4238.

[51] Tang Y, Han K, Guo J, et al. GhostNetv2: Enhance cheap operation with long-range attention[J]. Advances in Neural Information Processing Systems,2022,35:9969-9982.

[52] Han K, Wang Y, Tian Q, et al. Ghostnet: More features from cheap operations[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2020:1580-1589.

[53] Cai J, Gu S, Zhang L. Learning a deep single image contrast enhancer from multi-exposure images[J]. IEEE Transactions on Image Processing,2018,27(4):2049-2062.

[54] Guo X, Li Y, Ling H. LIME: Low-light image enhancement via illumination map estimation[J]. IEEE Transactions on image processing,2016,26(2):982-993.

[55] Ma L, Ma T, Liu R, et al. Toward fast, flexible, and robust low-light image enhancement[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2022:5637-5646.

[56] Shao X, Lyu Z, Li H, et al. Denet: an effective and lightweight real-time semantic segmentation network for coal flow monitoring[J]. Journal of Real-Time Image Processing,2025,22(1):1-16.

[57] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2016:770-778.

[58] Xu J, Xiong Z, Bhattacharyya S P. PIDNet: A real-time semantic segmentation network inspired by PID controllers[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2023:19529-19539.

[59] Shrivastava A, Gupta A, Girshick R. Training region-based object detectors with online hard example mining[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2016:761-769.

[60] Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2018:7132-7141.

[61] Guo C, Szemenyei M, Yi Y, et al. Sa-unet: Spatial attention u-net for retinal vessel segmentation[C]//2020 25th international conference on pattern recognition (ICPR). IEEE,2021:1236-1242.

[62] Wan C, Yu H, Li Z, et al. Swift parameter-free attention network for efficient super-resolution[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2024:6246-6256.

[63] Li H, Xiong P, Fan H, et al. Dfanet: Deep feature aggregation for real-time semantic segmentation[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2019:9522-9531.

[64] Si H, Zhang Z, Lv F, et al. Real-time semantic segmentation via multiply spatial fusion network[J]. arXiv preprint arXiv:1911.07217,2019.

[65] Yang M Y, Kumaar S, Lyu Y, et al. Real-time semantic segmentation with context aggregation network[J]. ISPRS journal of photogrammetry and remote sensing,2021,178:124-134.