| According to statistics,more than 1.5% of the world's population is unable to study and live as normal people because of visual obstacles.Image word recognition and voice play technology can provide reading help to some extent.Although the current market has been based on the Android terminal similar products,such as cloud file identification,OCR(Optical Character Recognition)text recognition,but these identification software on the image capture requirements are higher,often require clear text,the image can not tilt,the image only contains Text,etc.,otherwise it will not recognize or cause the recognition accuracy to be low.So these requirements for the presence of visually impaired people is not realistic.For this reason,this paper has developed a text image recognition software based on Android,and increased the function of voice playback,the user can obtain the text information through the auditory.The main work of this paper is as follows: Firstly,the text image tilt correction and text region clipping algorithm are proposed,and the information of the image image to be recognized is reduced by grayation,binarization,tilt correction and text area cropping.Preprocessing of text images.Then,based on google company optimized tesseract identification engine developed a text recognition function,and through training and expansion of the character library method to improve the accuracy of text recognition.Finally,based on the hand that TTS(Text To Speech)engine developed a voice playback function,the function can not only play the recognized text,and can be different gender,different volume,different speed to play.The validity of the image recognition and voice playback system based on Android platform is verified by the test of the system,and it is compared with the cloud file identification of one of the most widely used identification software in the market.Which is better when recognizing a text image that has a tilt or contains a non-text portion. |
PDF Full Text Request