CAJ | 학술논문

针对复杂多指针式仪表的读数自动识别难度大精度低的问题，提出了一种基于区域生长的指针式仪表自动识别方法。算法主要由基于区域生长的指针提取算法和基于 Hit-Miss 变换法击中直线或基于最小二乘法拟合直线的指针识别算法所组成。其中，区域生长所需的种子点通过基于差影法的模糊聚类自动选取。实验表明，基于区域生长的指针提取算法有效提取了指针特征，为 Hit-Miss 变换法和最小二乘法获得良好的指针识别精度奠定了重要基础。整个算法高效快速，能满足实时识别的应用需求。本文首次提出将基于区域生长的图像分割算法运用于指针式仪表识别领域中，丰富了指针式仪表识别的应用方法，获得了良好的识别效果。
침대복잡다지침식의표적독수자동식별난도대정도저적문제，제출료일충기우구역생장적지침식의표자동식별방법。산법주요유기우구역생장적지침제취산법화기우 Hit-Miss 변환법격중직선혹기우최소이승법의합직선적지침식별산법소조성。기중，구역생장소수적충자점통과기우차영법적모호취류자동선취。실험표명，기우구역생장적지침제취산법유효제취료지침특정，위 Hit-Miss 변환법화최소이승법획득량호적지침식별정도전정료중요기출。정개산법고효쾌속，능만족실시식별적응용수구。본문수차제출장기우구역생장적도상분할산법운용우지침식의표식별영역중，봉부료지침식의표식별적응용방법，획득료량호적식별효과。
Aiming at the difficulty of recogniseing complex multi-pointer instruments, this paper presents an automatic recognition system of analog measuring instruments based on region growing. The whole algorithm consists of pointer extraction algorithm based on region growing method and pointer recognition algorithm based on hit-miss transform method or least square method. Besides, the seeds needed by region growing can automatically obtained by fuzzy clustering based on difference image. Experiments show that, the pointer extraction algorithm based on region growing method effectively extract the features of pointers. And the extraction algorithm lays the foundation for the Hit-Miss transform method and the least square method to obtain good recognition accuracy. The algorithm is fast and efficient, can satisfy the requirements of real-time applications. This paper first applies region growing method of image segmentation to the field of analog measuring instruments recognition, enriches the method of analog measuring instruments recognition, achieved good recognition effect.