Research On Intelli Sense Technology Of Writing Behavior Of Chinese Characters With Wearable Device

In recent years,contact and contactless Intelli Sense devices widely exist in people's lives,and the research of behavior recognition based on Intelli Sense technology has been further developed.The contactless Intelli Sense devices mainly include cameras and wireless devices.However,contactless Intelli Sense technology based on camera is faced with the problems of high equipment cost,easy to be interfered by light conditions and high calculation energy consumption;contactless Intelli Sense technology based on sound and Wi Fi is faced with the problems of easy to be interfered by other signals and multi-path effects.All these restrict its large-scale popularization and application.Contact Intelli Sense technology is mainly based on wearable device,which has the characteristics of low cost,convenient deployment and not easy to be interfered by the external environment.And the smartwatch is welcomed by the majority of users because of its richer functions,which is more universal.With the popularization of Chinese,more and more people begin to learn Chinese characters.Writing Chinese characters correctly can help people improve their reading experience.Standardized writing behavior,including the pen-holding posture and the stroke order of Chinese characters,also directly affects the quality and speed of writing.For example,if you hold the pen with the wrong posture,it will lead to incorrect writing force,even myopia;if you write the Chinese character with wrong stroke order,it will destroy the balance structure of the Chinese character and reduce the writing speed.However,these problems are easy to be ignored for the beginners of Chinese characters,so the Intelli Sense technology of writing behavior based on Chinese characters has important research significance.Given the universality of smartwatches and the importance of writing behavior based on Chinese character,three key issues,namely Chinese character recognition,pen-holding posture recognition and stroke order recognition,are studied in this paper.(1)The handwritten Chinese character recognition method based on smartwatchThis paper realizes the recognition of the handwritten Chinese character based on built-in sensor of smartwatch for the first time.Firstly,by analyzing the handwritten Chinese character signal,an effective segmentation algorithm is proposed to improve the signal-to-noise ratio.Then,data enhancement method and deep convolution neural network method are used to build handwritten Chinese character recognition model.Finally,the factors that affect the performance of Chinese character recognition are analyzed.The experimental result shows that the precision of Chinese character recognition is 96%,which can recognize the handwriting Chinese characters well.(2)The pen-holding posture recognition method based on smartwatchThis paper realizes the recognition of the pen-holding posture based on built-in sensor of smartwatch for the first time.Firstly,by analyzing the handwritten signal,a new signal detection algorithm based on the pen-holding posture is proposed to improve the signal-to-noise ratio and extract Chinese strokes or English letters.Then,in view of the complex structure of Chinese characters and the low performance of the recognition model with characters strokes,this paper takes the combination of a vertical stroke and a horizontal stroke as the basic data of the pen-holding posture classification,and uses the deep convolution neural network to construct a two-level classifier to recognize the pen-holding posture,so as to further improve the recognition performance.Finally,the experiment result proves that the precision of pen-holding posture recognition is more than 98%,which meets the needs of people's daily use.(3)The stroke order of Chinese character recognition method based on smartwatchThis paper realizes the recognition of Chinese character stroke order based on built-in sensor of smartwatch for the first time.Firstly,by analyzing the handwritten signal,a new signal detection algorithm based on the stroke order of Chinese characters is designed to improve the signal-to-noise ratio,segment Chinese characters and extract handwritten signal.Then considering that it is difficult to distinguish the stroke order of Chinese characters only depending on the type and number of strokes,this paper uses the direction moving gesture to distinguish the relative order of the same strokes in the same Chinese character,so as to further improve the recognition performance of stroke order of Chinese characters.Finally,stroke and direction coding are designed and the Chinese character direction templates are to build.After all possible stroke order of Chinese character are compared,Chinese character stroke order is predicted based on the difference score.The experimental result also shows that the recognition accuracy of the model for stroke order of Chinese character is more than 83%,which has better recognition performance.