摘要
目前基于条纹投影轮廓术(FPP)的研究大多是通过重建整个场景或建立理想实验环境来实现目标分割的,但实际工业场景中周围环境复杂,直接使用FPP分割的相位含有大量无效区域,导致难以实现目标级测量。为此,提出了一种基于条纹投影轮廓术的相位有效区域分割方法,调制图和包裹相位图互为补充信息,同时作为输入,有效去除背景和阴影。通过自适应离散小波特征交互模块来学习和融合特征,采用离散小波变换(DWT)对特征进行分解,并利用边缘增强离散余弦变换(DCT)模块对特征进行解码,从而得到边缘清晰、结构明确的目标相位分割结果。定性和定量实验的结果表明所提方法比其他算法能更准确地分割出目标相位有效区域,验证了该方法在基于条纹投影轮廓术的目标级三维测量中的有效性。
At present,most of the researches based on fringe projection profilometry(FPP)achieve object segmentation by rebuilding the whole scene or establishing the ideal experimental environment,but in real industrial scenarios,the surrounding environment is complex,and the phase obtained by directly using FPP segmentation contains a large number of invalid regions,making it difficult to achieve target-level measurement.To this end,a phase-effective region segmentation method based on fringe projection profilometry is proposed.The modulation map and the wrapped phase map complement each other′s information and are used as inputs simultaneously to effectively remove the background and shadows.The features are learned and fused through the adaptive discrete wavelet feature interaction module.The features are decomposed by using the discrete wavelet transform(DWT),and the features are decoded by using the edge-enhanced discrete cosine transform(DCT)module,thereby obtaining the target phase segmentation results with clear edges and clear structure.The results of qualitative and quantitative experiments show that the proposed method can segment the effective area of the target phase more accurately than other algorithms,verifying the effectiveness of this method in the three-dimensional measurement of the target level based on fringe projection profilometry.
作者
刘建戈
吴柯啸
曹莉
张鹏宇
王伟业
LIU Jiange;WU Kexiao;CAO Li;ZHANG Pengyu;WANG Weiye(Huai′an Power Supply Branch of State Grid Jiangsu Electric Power Co.,Ltd.,Huai′an 210000,China)
出处
《电工技术》
2025年第13期81-86,96,共7页
Electric Engineering
