Research On Ballot Recognition Algorithm Based On Layout Understanding

The election is a widely used means of promoting democracy today, but democratic elections has a wide variety of ways, such as the traditional elections, the elections by showing hands, the electronic elections without paper and the electronic elections based on paper, in addition to a great variety of ballot formats, the voting way that can fully reflect voter's wishes and quickly count ballots has been pursued by people. Based on analyzing existing problems in the current researches, the electronic election based on the layout understanding is proposed, and the correlative principles, concepts and key technologies are studied as follows:First, the skew detection and correction method for a ballot based on straight line slope is advanced, and through analyzing the feature of ballots' batch scanning, the skew detection and correction method for batch ballots based on the adaptively nearest distance is proposed.Second, combining with the feature of traditional voting symbols, the symbol recognition method based on contour characteristics is studies, and about which some principles and correlative concepts are discussed. The method gives the solution to the recognition of the handwritten symbols. Third, combining with the feature of ballot layout and pre-defming the voting rule, a description method of layout structure based on undirected graph is presented, which makes location correct and batch counting fast according to a variety of ballot formats.Fourth, ballot recognition system adopts the electric election based on the layout understanding. The system is discussed and elementarily realized based on the Customer/Server mode, which confirms the effectiveness of the methods. Each function is programmed systematically.