Study on Small Target Extraction Algorithm Based on Deconvolution Normalization

Abstract

To improve the problem of small target information extraction in linear CCD detection system, the method of deconvolution is put forward to extract and analyze small target's image information under complex background. Through establishing a mathematical model to extract target information, the pure white template with the original background images that without target was deconvoluted, the complex background was transformed into a white plate template image, and then obtains the transfer function at the same time. After processing, the result shows that the background mainly becomes a piece of white board. While the target can be displayed clearly, a large number of gray-scale information of the target point can be retained at the same time. And the target can also recover the actual gray value of the point by using the reserved gray values, which has considerable research value for the judgment of small target and recovery of original target shape.

Keywords

• Linear CCD
• Image processing
• Small target extraction
• Transfer function
• Deconvolution
• Complex background.