Handwritten Symbols Recognition And Visualization Method Study In Questionnaire Atuomatic Statistical System

At present, in all kinds of social activities, batch recognition and automatic statisticsfor data forms by using image processing and pattern recognition techniques are importantto improve work efficiency and enhance processing accuracy. Although there are somespecial software systems based on OCR techniques, such as mail sorting system, banknotes and analysis system, automatic ballot statistical system, general data form automaticprocessing system for no fixed format constraints and limitations of a particular writing isstill relatively rare, and need to solve a number of related technical issues. At the same time,the current study for data visualization after recognizing image is not enough.Firstly, in aspect of checkbox recognition, this paper uses image processingtechnology and proposes an automatic recognition algorithm for checkbox, which candetermine whether checkbox is selected or not. First of all, based on the maximum tiltangle of runs of table lines, we detect and correct slanting form image, then apply theconnected domain labeling algorithm to rapidly extract the checkbox, finally extract itsstructure features to realize handwritten symbol recognition. We test100questionnaireimages about2000handwritten symbols and experiments show that this algorithm is better.Secondly, due to the amount of statistical data is large and multidimensional, thispaper discuss the visualization technology based on hyperbolic tree. To enhance itsvisibility, five interactive technologies (selectable marked-based interaction, exploratoryinteraction, summary/detail-type interaction, filter-type interaction, reconfigurableinteraction) are added to solve the label overlapping problem which is generated byhyperbolic tree in expressing large amounts of data.Finally, we conduct visualization subsystem design and implementation for thequestionnaire statistical system in aspect of the scientific quality. We use the conventionalmethod of data visualization and hyperbolic tree in the system and allow users to visualizethe desired results by a simple information interaction which can enhance the users to dynamically compare statistical data and reduce the workload analysis.