WIFI Electronic Endoscope Based On Android Smart Terminal

With the rapid development of China Mobile Internet,the portable and networked of therapeutic instruments has become hotspot in the field of telemedicine.In this paper,the main work from following aspects:design the hardware platform of WIFI electronic endoscope,design the Android terminal App and algorithm for endoscope software.According to the hardware platform,we first based on the existing front of endoscope which output analog video,concerned with the video conversion circuit which base on STK1160,and transmit the video to flush type circuit by USB interface to get the YUV/MJPEG video that can be identified by Linux development board,meanwhile,we build OpenWRT system and MJPG-Streamer in RT5350 WIFI embedded board,which can reduce the transmission bandwidth and compress the YUV/MJPEG video into multiframe JPEG to Android terminal,such as cellphones,tablets etc.,to achieve network transmission.In the aspect of software user interface design,this paper adopts the decorator pattern and singleton to design Inputstream object,and throw thread divide the stream into bitmaps,combined with the android widget Canvas and Paint to draw bitmaps on screen.Meanwhile,At the same time,this paper deeply discusses the coding structure of AVI video,and realizes the function of video playback and store as Mjpeg format AVI video.In order to prevent the conflict of display thread and record thread access to resources together,this paper uses cache queue to store the bitmap that has been drawn by display thread,record thread will visit cache queue and read the datas and write it into avi file.The user interface design also includes save endoscope image,browse image and image preview.In view of the algorithm of endoscope software,this paper realized the image enhancement and smoke removal,Android sdk had an API called ColorMatrix,which can input user defined parameters,thus modify the RGBA channel value of Bitmap,realize image enhancement.Smoke removal,which based on the dark channel prior model and mean filter,the computer vision model H(x)= F(x)t(x)+ A(1-t(x))under fog condition is solved,we estimate the parameter values of A and t(x),and restore the smoke image.At the same time,we through samplying the original picture and use window-shiffting model on mean filter algorithm to optimize the smoke removal algorithm,and finally realize frame rate of 20fps under this algorithm,which can be used in real-time condition.