Research On Real-Time Distortion Correction Hardware System Of Color Image For Medical Electronic Endoscope

Medical Electronic Endoscope (MEE) is a widely used medical instrument,through which physicians can directly observe the form and pathological changes ofinside organs and diagnose illness easily. Because wide-field lens are adopted in MEE,output images have serious distortion. Software, limited by computer's calculatingspeed, can only correct the distortion of static images but not the distortion ofdynamic images. So we use the high speed character of very large scale integration(VLSI) to reach real-time need. From the point of digital image processing, image distortion correction isactually the problem of image renewing, which includes spatial correction andgray-level correction. This paper is based on the dot-array plate correction method.Standard dot-array plate is placed in front of MEE, after whose optical system thedistorted dot-array plate image is received. In comparison the standard dot-array plateimage with the distorted one, the distortion function can be found to implement spatialcorrection. And zero-class interpolation is selected for correction of gray level in viewof the speed of hardware access. The method of hardware distortion correction is that the spatial correction dataare calculated first under software circumstance, then converted into the address ofhardware memorizer and stored in the hardware look-up table, through which data canbe read out to correct gray level. In order to minimize the amount of hardware look-uptable and minish the volume of hardware system, hardware correction process shouldbe imitated by software to insure the quality of corrected images. The function of hardware system includes decoding of video signals, correctionof distorted images and encoding of corrected digital images. SAA7111, A/D videochip of Philips Co. Ltd., is selected as the decoder to convert video into 16 bit YUVdigital images for correction; Bt864, D/A video chip of Conexant Co. Ltd., as theencoder to output the video of corrected images; SRAM as the memorizers to storeinput and output images; EPROM as the look-up table; FPGA as the video controlcenter to implement control of other chips. FPGA, acting as the control center, is utilized to take charge of decoding,correcting and encoding, so program design of FPGA is the most important part ofthis paper. In accordance with the function of each chip, we introduce the generationmethod of outside logical control in detail and present the related waveform figuresand the debugging result of the whole hardware system. At the end of this paper, we analyze the error source of distortion correctionhardware system, compare the effect of distortion correction, summarize thecharacters of system and put forward the improvement way. This hardware systemworks in PAL; field frequency is 50Hz; size of correcting image is 702x576 pixels;input and output are both standard video signals so that corrected images can bedisplayed on monitor directly. In distortion correction aspect, the distorted image'srelative distortion is –19.2% and the corrected image's relative distortion is -0.7%, sowe can conclude that the hardware system effectively reduce the distortion of optical IIsystem. In real-time aspect, delay time between distorted images' input and correctedimages' output is 40ms, which human's eyes cannot be aware of at all, so thehardware system reaches the demand of real-time character.