In Japan, it is reported about 90% of the persons with visual impairment are the persons with low visual capability. However, the database of the low visual capability and the standard about the easily comprehensible visual information presentation are not existed yet. Therefore, the categories of colour similarity of the persons with low visual capability are measured. However, measuring various colour combinations is placing a burden on the object person. Therefore, in this research, development of the categories of colour similarity estimation algorithm is proposed. Applying the generalization capability of the artificial neural network, the distribution of the categories of colour similarity will be obtained from the limited colour combination measures.