中文摘要：

大背景中微小目标检测的一个重要问题是可疑目标区域的确定。以棉花中污杂物的机器视觉检测为背景,提出了一种可疑目标区域算法：首先用DCT算法提取图像的频率特征,然后用基于ADABOOST的算法识别棉花图像中是否有污染物。实验结果表明算法达到了较高的识别精度。

英文摘要：

One of the key points in a machine vision inspection of small targets in a large background is how to locate the suspicious target regions.ADABOOST was used for detecting suspicious regions in cotton images.Firstly,discrete cosine transform（DCT） was adopted to compress the images and extract their main features.Then an algorithm based ADABOOST was used to identify suspicious target area.Experimental results indicate that the algorithm is effective.