Abstract: Traditional edge detection operators have defects when extracting overhead transmission lines from images. The common issue is that the extracted image of transmission lines is incoherent and inaccurate. In order to solve this problem, an improved Canny edge detection method is proposed. Aiming at solving the problem of the single template of the traditional Canny operator, a 4-direction improved template is given and the larger value of the tilt angle is selected for the gradient calculation. Aiming at overcoming the limitation of threshold selection, an adaptive threshold selection method of Otsu algorithm is proposed by combining the intercept method with the artificial bee colony algorithm. The experimental results show that compared with the improved Otsu-PSO threshold selection algorithm, this method has improved anti-noise ability and accuracy by 9.38% and 1.7% respectively and the single-edge response effect is also improved, which has a certain significance for enhancing the transmission lines edge detection technology.