复杂背景下图像中弱小目标的分割和提取算法

针对有些图像中的目标与背景之间存在区域边界往往不明显，加上噪声的影响，目标边界具有一定的模糊性，有时还存在目标与背景面积对比差别很大等问题，提出了一种基于图像边缘增强的自适应最佳阈值图像分割方法，有利于将弱小目标从复杂背景中精确地分割和提取。该算法将闽值法和其它方法进行结合，有效地解决了目标边缘模糊图像阈值选取困难这一问题；采用图像切割技术，有效解决了弱小目标在大背景下分割难的问题；将类别方差法和迭代算法相结合建立数学模型，不但有利于自适应最佳闻值的选取，而且有效地减少噪声对最佳自适应阈值的影响，提高了图像的分割效果。

A Segmentation and Distilling Algorithm for Small Objects of Image in Complicated Background

The boundary between the object and the background was not visible in some images.The edge of object was fuzzy when the noise interferes.Sometimes there was much difference in the area between the object and the background.An adaptive best threshold value image segmentation algorithm based on image enhancement was given,which was effective in segmentation and distilling the small object of an image.The algorithm effectively solved the problem which the threshold was difficult to choose for the fuzzy image by combining threshold method with other method.The problem that the small object was difficult to segment from the large background was solved effectively by adopting image incision technology.A mathematic model was built by combining sort square difference method with iterative algorithm.It was helpful to select adaptive best threshold value and can decrease the influence of the noise.The segmentation effect for image was improved.