In recent decades, due to the use of computer technology, the results of research in a particular area of science and technique are accumulated very quickly. In the process of creating a database of optical images and knowledge base is appeared the task of classification method absence. Without such a method the creating the structure of data warehouse is difficult to solve. The classification of the tasks and the obtained results requires considerable attention in order to ensure their rapid and accurate search and further use. Classification is one of the most important tasks of large database analysis used in Data Mining. Assessment of classification accuracy can be performed using cross-checking or a test set. The proposed method is suitable for use when the accuracy of the test set classification meets the established requirements. In the field of optics and laser physics dealt with the problem, which usually is grouped by the next attributes: the initial field pattern, the type of the optical system, the characteristics of the obtained spatial distribution of light as the number of maxima, minima, zeros of intensity and other significant objects. The question of classification of the tasks themselves was considered in the paper for the first time. The method of classification of problems in optics and laser physics is developed. It is applied to the problems of diffraction, interference and observation of microobjects. The classification is carried out with the help of mathematical methods, a formal description of the objects and the decision trees method. The 5 classes was chosen: "meaning the task", "solution method", "appointment", "complexity", "way of expression." The attributes of the problem classification are "visual-optical", "optical", "computational", "experimental", "qualitative", "fundamental", "applied", "simple", "complex", "textual", "numerical", "graphical". The scheme of problems in optics and laser physics classification is shown. Decision-making to determine the attributes of the problem is carried out using the decision trees method. Application of the results in search engines of big databases and pattern recognition problems is discussed.