Indicator State Recognition Based on HSV Feature Transformation
Rongrong Wu, Wei Zhang, Hong Chen, Jian Jiao · 2020
To improve the recognition accuracy of substation inspection robots for densely arranged and colorful indicator lights, a method of indicator state recognition based on color clustering is proposed. Firstly, a holomorphic filter and calibration are adopted to overcome the influence of illumination and the shooting angle. Then, the s-component clustering segmentation method is used to locate the indicator, and row and column projections are carried out on the positioning result to extract the projection curve. Finally, the peak values are used as the basis for dividing the row and column, and the identification of the indicator array of multiple rows and columns is converted into a single indicator unit in a small area. Experimental test results showed that the recognition rate of this method could reach 98.5%.