The invention relates to a method for losslessly evaluating ozone sensitivity of a plant through leaves, belongs to the field of atmospheric pollution biological monitoring, and the method can be applied to lossless and continuous determination of injury symptoms of ozone sensitive plants such as soybeans, poplar trees, silver birches, tilia amurensis and quercus liaotungensis. The method comprises the following steps: classifying damage types according to a leaf surface image, then segmenting the leaf surface image, extracting and analyzing characteristics such as symptom shape and area size, establishing an extraction method based on an ISODATA algorithm in a Lab color mode, obtaining a spot rate, and then performing damage degree grading (figure 1). The method has the advantages that compared with an existing naked eye observation or destructive sampling method, the method can analyze the damaged area and proportion of the leaf surface without damage, continuous monitoring and accurate grading are achieved, the problem that the damage degree cannot be quantitatively compared due to the fact that ozone damage symptoms of different plants are different can be solved, the plant ozone damage can be simply and directly judged, and the method has the advantages of rapidness, no damage, continuous observation and the like.