Considering the complicated and variable background of transmission line image, the single method of image processing is difficult to effectively deal with various background types of transmission line images, thus a method of image classification and feature extraction based on color space variable is proposed. Firstly, according to the relationship between the variable value of each color space and the image feature of the transmission line image, the image is classified. Then, the appropriate filtering method is selected according to the different image features. The noise is removed by combining two filtering, and the adaptive histogram segmentation equalization is used to enhance the image contrast. The parameters of the canny edge detection are determined by linear transformation of the threshold obtained by the Otsu algorithm to extract transmission line edge. Finally, according to the shape feature of transmission line and the probability Hough linear transformation and morphological operation, a method of edge optimization is proposed and the non-edge of transmission line is removed better. The results show that the method can effectively deal with various background types of images, providing a new idea and method for the intelligent processing of transmission line images. This work is supported by Science and Technology Program of Heilongjiang Province (No. GZ11A213).