A Text Recognition Method for Infrared Images of Electrical Equipment in Substations

As for the operation and maintenance of substations, infrared detection is an effective way of monitoring states of electrical equipment. However, the number of images taking by infrared thermometer increases with more and more equipment putting into operation, which aggravates the burden of recording temperature from infrared images of equipment. To decrease workload and improve work efficiency of operation and maintenance staff, an image recognition method for infrared images based on HSV (Hue, Saturation, Value) color model, CTPN (Connectionist Text Proposal Network) and STRF (Scene Text Recognition Framework) is proposed. Images are processed by HSV to extract the color of text. CTPN is used to transform multi-line text images into single-line text images. STRF is trained to recognize text on the images. Simulations are carried on to demonstrate effectiveness of the proposed method.