基于显著性先验和金字塔转换机制的 Trans-Sal Det小目标油井检测模型

赵梓翔; 李佳慧; 步贤业; 穆树娟; 隋杨

doi:10.3969/j.issn.2095-4107.2026.02.008

东北石油大学学报 ›› 2026, Vol. 50 ›› Issue (2) : 109 -122. DOI: 10.3969/j.issn.2095-4107.2026.02.008

计算机与自动化工程

基于显著性先验和金字塔转换机制的 Trans-Sal Det小目标油井检测模型

赵梓翔 ¹^,²^,³ ,
李佳慧 ²^,⁴^,⁵ ,
步贤业 ²^,³^,⁴ ,
穆树娟 ²^,³^,⁴ ,
隋杨 ⁶

作者信息 +

Trans-Sal Det small object detection model based on saliency prior and pyramid transformation mechanism

Zixiang Zhao ¹^,²^,³ ,
Jiahui Li ²^,⁴^,⁵ ,
Xianye Bu ²^,³^,⁴ ,
Shujuan Mu ²^,³^,⁴ ,
Yang Sui ⁶

Author information +

文章历史 +

PDF (9817K)

摘要

为解决复杂地表环境与背景干扰条件下的油田遥感影像中油井小目标难以准确识别的问题，提出一种融合显著性先验与金字塔转换机制的 Trans-Sal Det 小目标油井检测模型。构建基于 Tiny-U-Net 的显著性生成模块，获取高分辨率显著性图，突出潜在油井区域并抑制冗余背景信息；将显著性图与原始遥感影像进行通道级拼接，输入金字塔式 Transformer 编码器，采用多尺度窗口自注意力机制，实现跨层特征建模，有效融合底层细节信息与高层语义特征；引入交叉尺度特征融合策略，增强模型对不同尺度油井目标的表征能力。在典型油田遥感数据集上开展对比与消融实验，对模型性能进行验证。结果表明：Trans-Sal Det 小目标油井检测模型在复杂背景、遮挡及低对比度场景下表现出更优的小目标检测能力，目标召回率与检测精度显著优于其他方法的。引入显著性先验可以有效引导注意力聚焦关键区域，提升 Transformer对小目标的感知能力。该结果为油田自动化遥感检测与智能监测提供高效和可行的技术路径。

Abstract

To address the difficulty of accurately detecting small oil well targets in oilfield remote sensing images under complex surface environments and background interference, this paper proposes a TransSal Det small object detection model that integrates saliency priors with a pyramid transformation mechanism. A saliency generation module based on Tiny-U-Net is constructed to produce high-resolution saliency maps, highlighting potential oil well regions while suppressing redundant background information. The saliency maps are concatenated with the original remote sensing images at the channel level and fed into a pyramid Transformer encoder. Through a multi-scale window self-attention mechanism, cross-layer feature modeling is achieved, effectively integrating low-level detailed features with high-level semantic information. A cross-scale feature fusion strategy is introduced to enhance the model's representation capability for oil well targets of different scales. Comparative and ablation experiments are conducted on typical oilfield remote sensing datasets to systematically validate the model performance. The results demonstrate that Trans-Sal Det exhibits superior small object detection performance in complex backgrounds, occlusion, and low-contrast scenarios, with significantly improved recall and detection accuracy compared to mainstream methods. The introduction of saliency priors effectively guides attention to focus on key regions, enhancing the Transformer's perception of small targets. The proposed method provides an efficient and feasible technical approach for automated oilfield remote sensing detection and intelligent monitoring, with strong potential for practical applications.

关键词

显著性先验 / 金字塔转换机制 / 交叉尺度融合 / 油井检测 / 动态显著性 / Transformer / 小目标检测

Key words

saliency prior / pyramid transformation mechanism / cross-scale feature fusion / oil well detection / dynamic saliency / Transformer / small object detection

引用本文

引用格式 ▾

赵梓翔,李佳慧,步贤业,穆树娟,隋杨. 基于显著性先验和金字塔转换机制的 Trans-Sal Det小目标油井检测模型[J]. 东北石油大学学报, 2026, 50(2): 109-122 DOI:10.3969/j.issn.2095-4107.2026.02.008

登录浏览全文

4963

注册一个新账户忘记密码

参考文献

原文顺序 | 出版日期 | 本文引用

[1]	Gui S, Song S, Qin R, et al. Remote sensing object detection in the deep learning era:a review[J]. Remote Sensing, 2024, 16(2): 28.

[2]	Teodosio L, Emanuele C, Irina C, et al. A MODIS-based robust satellite technique(RST)for timely detection of oil spilled areas[J]. Remote Sensing, 2017, 9(2):128.

[3]	江濒, 张微, 陈静, 等. 基于改进 YOLOv8 算法的去雾遥感图像电力目标检测[J]. 遥感技术与应用, 2025, 40(5):1243-1254.

[4]	Jiang Hao, Zhang Wei, Chen Jing, et al. Power target detection based on improved YOLOv8 algorithm in dehazing remote sensing image[J]. Remote Sensing Technology and Application, 2025, 40(5):1243-1254.

[5]	陈维力, 潘永帅, 范坤宇, 等. 塔里木盆地库车坳陷博孜一大北地区差异成藏过程及控藏因素[J]. 东北石油大学学报, 2025, 49(1): 61-76.

[6]	Chen Weili, Pan Yongshuai, Fan Kunyu, et al. Pore structure differential accumulation process and reservoir controlling factors in Bo- zi-Dabei Area,Kuqa Depression,Tarim Basin[J]. Journal of Northeast Petroleum University, 2025, 49(1):61-76.

[7]	彭先锋, 胡笑非, 张烨毓, 等. 苏里格气田山 1 段储层致密化成因及控制因素[J]. 东北石油大学学报, 2018, 42(1):60-67.

[8]	Peng Xianfeng, Hu Xiaofei, Zhang Yeyu, et al. Densification and controlling factors of tight reservoirs in Shan 1 Formation of Sulige Gas Field[J]. Journal of Northeast Petroleum University, 2018, 42(1):60-67.

[9]	Ral A, Kim J M. A novel pipeline leak detection approach independent of prior failure information[J]. Measurement, 2021, 167: 108284.

[10]	Wang C, Wang Z, Dong H, et al. Fusionformer:a novel adversarial transformer utilizing fusion attention for multivariate anomaly detection[J]. IEEE Transactions on Neural Networks and Learning Systems, 2025, 36(8):14479-14492.

[11]	Mahmood Y, Afrin T, Huang Y, et al. Sustainable development for oil and gas infrastructure from risk,reliability,and resilience perspectives[J]. Sustainability, 2023, 15(6):4953.

[12]	石颖, 李莹, 王维红, 等. 线性 Radon 变换噪音压制法及其在古龙断陷中的应用[J]. 东北石油大学学报, 2012, 36(4):116-120.

[13]	Shi Ying, Li Ying, Wang Weihong, et al. Approach of linear noise suppression using Radon transform and its application in Gulona fault depression[J]. Journal of Northeast Petroleum University, 2012, 36(4):116-120.

[14]	Cordes E E, Jones D O B, Schlacher T A, et al. Environmental impacts of the deep-water oil and gas industry:a review to guide management strategies[J]. Frontiers in Environmental Science, 2016, 4:58.

[15]	Vekeen S T, Balogun A L. Advances in remote sensing technology,machine learning and deep learning for marine oil spill detection, prediction and vulnerability assessment[J]. Remote Sensing, 2020, 12(20):3416.

[16]	李睿. 油气管道内检测技术与数据分析方法发展现状及展望[J]. 油气储运, 2024, 43(3):241-256.

[17]	Li Rui. Current progress and prospects of in-line inspection techniques and data analysis methods for oil and gas pipelines[J]. Oil & Gas Storage and Transportation, 2024, 43(3):241-256.

[18]	高胜, 王妍, 任永良, 等. 大型复杂油田注水系统优化运行关键技术与智能化展望[J]. 东北石油大学学报, 2020, 44(4):91-98.

[19]	Gao Sheng, Wang Yan, Ren Yongliang, et al. Key technologies and intelligent prospect for optimized operation of large complex wa- ter injection systems in oilfields[J]. Journal of Northeast Petroleum University, 2020, 44(4):91-98.

[20]	Brown K E, Lea J F. Nodal systems analysis of oil and gas wells[J]. Journal of Petroleum Technology, 1985, 37(10):1751-1763.

[21]	苏健, 叶文强. 复杂环境下的小目标交通标志检测[J]. 计算机系统应用, 2025, 34(11):202-211.

[22]	Su Jian, Ye Wenqiang. Traffic sign detection for small objects in complex environment[J]. Computer Systems & Applications, 2025, 34(11):202-211.

[23]	Li Z, Wang Y, Zhang N, et al. Deep learning-based object detection techniques for remote sensing images:a survey[J]. Remote Sensing, 2022, 14(10):2385.

[24]	Yang Y, Wang J, Liao J, et al. Abundance and diversity of soil petroleum hydrocarbon-degrading microbial communities in oil explo- ring areas[J]. Applied Microbiology and Biotechnology, 2015, 99:1935-1946.

[25]	Zhou X, Chang N B, Li S. Applications of SAR interferometry in earth and environmental science research[J]. Sensors, 2009, 9(3): 1876-1912.

[26]	王晓丽, 邓达康, 孟祥龙, 等. 基于领域本体的油气勘探开发知识获取模式及实现[J]. 东北石油大学学报, 2016, 40(4):74-79.

[27]	Wang Xiaoli, Deng Dakang, Meng Xianglong, et al. Oil and gas exploration and production knowledge processing mode and imple- mentation based on domain ontology[J]. Journal of Northeast Petroleum University, 2016, 40(4):74-79.

[28]	Behera D K, Pujar G S, Kumar R, et al. A comprehensive approach towards enhancing land use land cover classification through ma- chine learning and object-based image analysis[J]. Journal of the Indian Society of Remote Sensing, 2025, 53(3):731-749.

[29]	Yang C, Huang Z, Wang N. QueryDet:cascaded sparse query for accelerating high-resolution small object detection[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. New Orleans: IEEE,2022:13668-13677.