Research On Signal Processing Technique Of CMOS Photoelectric Autocollimator

With the development of the science technology, it needs a higher request to miniaturization, high speed and anti-jamming of the device. The CMOS imager is considered as an advanced photoelectric converter device, its integration, speed, power consumption and cost is better than CCD and gets many scholar's attention. At present, in the photoelectric autocollimator field, CMOS imager is on the exploring phase and has not a ripe product in and out of the country. In this paper , the author focuses on the photoelectric autocollimator based on CMOS imager , its emphasis is the digital signal processing .At first the author analyzes the principle of the photoelectric autocollimator,especially the digital signal processing .Then considering the low quality imaging of the CMOS imager and the higher request for the photoelectric autocollimator to the SNR., the author proposes the way that restrains the Gaussian white noise and the impulse noise by median filtering and Gaussian mixture model denoising algorithm based on FPGA, it is necessary to pre-process the adopting noised image and improve the SNR.Matlab is used for algorithm simulation to the effect of this mixture model denoising at first. By experimentation the author gets a summery: the effect of passing Guassian filter after median filter is better than the effect of passing median filter after Guassian filter ,at the same time it is proved that is a better way to restrict the Gaussian white noise and the impulse noise. Then this algorithm is realized in FPGA using VHDL, and it improves the quantity and SNR of the image.After FPGA gets the aim information, the information is transferred to the DSP, then using DSP calculates the offset and the misalignment angle. Finally DSP transfers the result to the FPGA and the result is showed on TFT screen in time .On the other hand , the result is transferred to PC computer using serial port in order to data processing on PC computer. The experimenting result indicates: the mixture model denoising algorithm using FPGA can improve the SNR of the CMOS imager and the misalignment angle measuring precision of the photoelectric autocollimator. It is validated the reliability of the data communication between DSP and PC computer, the data measuring by PC is the same as the result using the JTAG scan , it is realized the complex measuring using the DSP combining the PC computer.