Nowadays,most of the surveillance systems on the market use infrared cameras.The Charge Coupled Device(CCD)and Complementary Metal Oxide Semiconductor(CMOS)sensor not only receive visible light but also near-infrared light.To deal with the crosstalk problems of three channels caused by the near-infrared light on R,G,B,the methods used in the current surveillance field is to add an electric control device called IR-CUT switcher in an optical imaging system.Therefore,the infrared light is filtered out during the day,and the infrared light compensation function is used at night to achieve day and night monitoring.However,there will be oscillations in the back and forth switch in the morning or in the evening,which may cause problems such as inadequacy of filter switching,high failure rate,and the decline of life and so on.This paper proposes a four-band image color correction algorithm based on the infrared crosstalk measurement of the camera module.The algorithm adopts the idea of polynomial regression to construct the crosstalk matrix,uses the least square method to fit the optimal solution crosstalk matrix,replaces the mechanical switching device with the algorithm,and consequently avoids the mechanical failure.The method utilizes the numeric value on the color card to get the correction matrix.Then,the de-crosstalk array is used to calibrated the four-band images under the same light source condition,so as to achieve the purpose of filtering the crosstalk of infrared light.The experimental results showed that the method had good correction effect in the indoor environment and outdoor setting to some degree.Moreover,the crosstalk matrix needs only one-time calibration,and could make a color correction for four different images under the same source condition,which made a useful exploration for further restoration of four-band color pictures. |