基于图像处理和BP神经网络的水位识别研究

Research on Water-Level Recognition Based on Image Processing and BP Artificial Neural Network Technology

为了能够自适应地识别水位,针对目前水尺图像类水位识别软件中需要人工设定水尺总量程参数的弊端,提出了一种无需在软件中预设水尺总量程的水位监测方法。该方法通过训练好的BP人工神经网络智能识别水尺总量程,通过图像处理技术提取水尺刻度,最后利用水尺总量程和水尺刻度之间的数学关系计算出水位值。实例应用结果表明:该方法能够有效地识别水尺总量程和水位值。

In order to adaptively identify the water level from water-level ruler image, a novel method which detects the water level without manually presetting the water-level ruler gauge was proposed.The water-level ruler gauge was identified by trained BP Artificial Neural Network and the water-level ruler scale was extracted by image processing.Then, the water level was obtained by making full use of the mathematical relationship between the water-level ruler gauge and the scale.Experimental results show that the proposed method can identify the water-level ruler gauge and water level effectively.