Abstract: Tow-point Non-uniformity Correction (NUC) based on blackbody is not suitable to correct actual sky background. To overcome this problem, a new two-point correction method based on gradient sky background was proposed. First, the fixed positions of bind pixels are replaced by the pixels in neighborhood. Then, the two-point images are captured at two different pitch angles in the pure sky by infrared focal plane array detector; and gain and offset coefficients are obtained from two-point correction. Finally, the two correction coefficients are modified to remove a few remaining bas points. Results show that this method has good performance. In the second place, a new bilateral filtering and gradient template algorithm is proposed to detect dim point target after correction. First, median filter is used to remove remaining single noise. Then, convolution with the gradient template is made. Dim point target can be detected after adjusting image grayscale to the certain threshold interval. Results show that the proposed gradient template algorithm has advantages to suppress complex background. It improves Signal-to-noise Ratio (SNR), and has the preference of low complexity and easy implementation in real-time system.

Key words: information processing technology, gradient sky background, non-uniformity correction, gradient template algorithm, dim point target