The paper discusses the progress in the automation of the detection of flare stars through astronomical observations with plates obtained by the multi-exposure “chain” method. In spite of the fact that the astronomical plate observations are considered obsolete in modern astronomy, the computer processing of these images is an actual problem, due to the considerable number of wide-field plates in our country, in Europe and throughout the world. The solution of the problem is implemented in a sequence of three stages: (i) detection of the form of the chains, (ii) location of the chains in the input plate image and (iii) detection of chains representing “flare stars”. The paper is focused on the second stage, the implementation of which requires the use of Hough transform (HT). A theoretical and experimental analysis of the suggested approach connected with HT is presented. The “bottlenecks” in the development are discussed on the basis of a comparative analysis with a competitive approach - the “cumulative” statistic approach to chains in the given image. At the end, a combined approach is offered, uniting the various advantages, but again based on HT. Some experimental results and illustrations are presented. The perspectives for future work, connected with an implementation of a whole software system designed to solve the problem, are described.
