Research Of FPN Correction Method And System Implementation Based On TDI-CMOS Image Sensor

The ordinary line-array image sensor which has one-dimensional pixel array structure,can capture two-dimension images by scanning motion.The quality of the images captured by the sensor is limited by the illumination intensity and relative velocity between sensor and the imaging object.Time delay integration(TDI)is a kind of technology which uses the two-dimension pixel array structure to realize the function of line-array image sensor.Signals from the same imaging object can be sampled and accumulated by pixels in different rows and the equivalent exposure time increases.Therefore the signal-to-noise ratio can be improved effectively.Besides,the CMOS devices have many potential advantages,such as low power consumption,process accessibility,high integrated density,etc.The TDI-CMOS image sensor(TDI-CIS)is attracting more and more attention.With the study research of TDI-CIS,some new problems have arisen.For example,there are some stripes which is caused by the spatial distribution characteristics of fixed-pattern noise(FPN)in the output image.It seriously affects the image quality.In order to eliminate the FPN in the output image of TDI-CIS as well as improved image quality,the working principle of TDI-CIS is studied firstly.Then the source,characteristics,expression forms and mathematical model of FPN are analyzed.And on this basis,a FPN correction method based on moment matching is proposed.Firstly,the original image is processed with spatial correlation filter(SCF),and next the filtered image is processed by improved moment matching method in the along-track direction.Image with row FPN correction can be got.Secondly,the image with row FPN correction is processed with bilateral filter,and next the filtered image is processed by improved moment matching method in the across-track direction.Finally,image with FPN correction can be got.In order to validate the effectiveness of the algorithm,an imaging system is designed based on the TDI-CIS with on-chip analog accumulator,field-programmable gate array and USB development board.External connection circuit for imaging system and internal functional circuit for SOPC are studied and designed.Data of images captured by TDI-CIS are transmitted to personal computer firstly.Then the FPN correction process is conducted with MATLAB.Experimental results show that,after correcting the FPN in the image captured under uniform illumination,the standarddeviation of row-mean vector decreases from 5.6761 LSB to 0.1948 LSB,while the standard-deviation of column-mean vector decreases from 15.2005 LSB to 13.1949 LSB.Besides,for different images captured by different TDI-CISs,the average decrease of standard-deviation of row-mean vector and standard-deviation of column-mean vector is 5.4922/2.0357 LSB respectively.In addition,the change trend of both the row-mean and column-mean curve of the corrected image is almost consistent with that of original image as well as smoother than that of original image.Comparative experimental results indicate that the proposed method can correct the FPNs of different TDI-CISs effectively while maintaining image details.