To solve the problem of the lack of effective evaluation of environmental pre-judgment by the intelligent inspection robot of the substation, the meter reading in the fog environment is considered the research object, and a support vector regression (SVR)-based meter reading environmental pre-judgment model is proposed. The proposed model uses discrete cosine transform (DCT) frequency domain and spatial structure features based on the local binary pattern rotation invariant operator to reflect the fog density. Besides, it uses the statistical features of the depth image to reflect the distance and SVR to train and fit all image features. Thus, the influence of fog density and distance on the meter reading is comprehensively considered, and the discrimination accuracy is improved. The proposed algorithm is tested on the collected image database and compared with the existing algorithms. The experimental results show that the addition of the depth map feature significantly improves the performance of each algorithm compared with the absence of the depth map feature. It effectively shows the influence of the screen ratio on the meter reading. Compared with other related algorithms, the performance of the proposed algorithm is optimal. The proposed algorithm can effectively solve the problem of environmental pre-judgment.