基于稀疏表示及奇异值分解的水位检测系统设计

Water Level Detection System Based on Sparse Representation and Singular Value Decomposition

水位检测系统中摄像机采集到的水尺图像可能会存在刻度模糊、局部缺失、小角度旋转等情况,为了克服以上问题和提高在缺乏训练样本时的检测识别效果,提出了一种奇异值分解与稀疏表示相结合的水位检测识别算法,能快速有效测量出水位值。首先在Hue Saturation value(HSV)空间下通过形态学预处理操作提取出水尺目标区域,再运用图像局部及整体的奇异值作为图像特征向量,接着采用稀疏表示的方法进行分类,最后根据分类结果计算出水位值。实验结果表明,该算法对水尺刻度不清或旋转等情况具有较强鲁棒性,能对较远距离情况下拍摄的水尺图片进行准确有效的水位识别。

In order to overcome the situation of scale blur,local missing,small angle rotation in the traditional water level detection,and in order to reduce the feature vector dimension,In this paper,a new method based on singular value decomposition and sparse representation is proposed,which can be used to measure water level quickly and effectively.The target area of water ruler is extracted by morphological preprocessing operation in HSV space,the singular values of local and whole image are used as image features,then the sparse representation method is used to classify,and the water level value is calculated according to the classification results.The experimental results show that it has strong robustness to the situation such as the scale or rotation of the water scale,and can accurately and effectively identify the water level of the water scale image obtained at a longer distance.