Based on a set of image processing techniques, an automatic approach for reading resistance of a carbon film or metal film resistor through its color bands is presented. light source is investigated and light condition under controlled is recommended to help the extraction and identification of color bands. Resistor can be put at any orientation for recognition. Given a resistor image, the proposed approach is mainly composed of segmentation and horizontal alignment of resistor's body, extraction of color bands, as well as classification and identification of color bands. Verification of flipping over sequence of extracted color bands is also involved in our approach. K-Nearest Neighbor (K-NN) classification is used to train and classify both main body color and extracted band's colors. The correctness of extracting and counting color bands is 96%. Correctness of reading resistor value is 92%. The results confirm the feasibility of the proposed approach.