This paper presents a different perspective on the evidence theory and Dijkstra's algorithm. In this paper, the algorithm will be used in the further analysis to additional paths between nodes can be found in the maritime sector. In many cases, the best solution for a single criterion is not sufficient. It would be the search for more effective solutions of the starting point to use for subsequent analysis or decision of the captain of the ship. With innovative thinking mechanisms, it is possible to create a decision support system based on known Dijkstra's algorithm. Reasoning mechanism is based on the theory of evidence. The intervals were used, where interval means a set of real numbers with the property that any number that lies between two numbers in the set is also included in the set.

Accuracy of character recognition is influenced mostly by the segmentation phase of the optical character recognition. The segmentation of touching and overlapping characters which occurs often in handwritten text of south Indian languages has been remained as a challenging problem. This paper proposes an iterative technique performing segmentation of touching/overlapping Telugu characters which are commonly found in handwritten words Telugu application form documents. The document is initially segmented into lines and then into words using horizontal and vertical projection profiles. The vertical projection profile of each word is used to segment the word into characters which are characterized into valid character blocks and invalid character blocks. Invalid character blocks are considered as blocks consisting of touching or overlapping characters with in the word. The invalid character blocks and valid character blocks are distinguished by applying correlation analysis between segmented blocks and a database of valid character classes. Further the identified invalid character blocks alone are processed separately by an iterative split analysis technique which employs convenience sampling methodology for the identification of candidate segmentation points between characters. The experimental results are adequate and consistent for the processing of invalid character blocks consisting of one or multiple touching or overlapping character blocks found in handwritten words of application form images. The average accuracy achieved at various sampling percentages in iterative split analysis process is found to be 91.17%.