中文摘要：

单位面积麦穗数是小麦产量预测中的一个重要参数。在应用图像处理技术识别麦穗的个数研究中，提出了一种小麦颜色信息和AdaBoost算法相结合的麦穗检测方法。先用麦穗颜色分割法排除非小麦区域的干扰，然后用AdaBoost算法训练的分类器对麦穗区域进行定位识别。该方法在保证检测率的同时，大大减少了目标区域的误检率，提高了小麦检测准确率。试验证明所用方法能有效去除叶片和秆茎的干扰，对于角度倾斜有一定鲁棒性。在随机抽取的100个样本中检测率为88．7％，有很强的识别能力。

英文摘要：

The number of wheat per unit area is an important parameter in the wheat production forecast. There- fore,in order to apply image processing techniques to identify the number of wheat,we proposed a method of spike detection that combined wheat color information and AdaBoost algorithm. First we excluded the interference of non- wheat area using wheat color segmentation method,and then used the AdaBoost algorithm trained classifier to locate and identify the wheat area. In this method, the detection rate has been guaranteed, at the same time, the rate of false detection of the target area has been gready reduced,and the detection accuracy of wheat has been improved. The experiment proved that the method used herein could effectively remove the interference of the blades and stems, and it has a certain robustness inclined at an angle. It has a strong identification capability,and more than 88.7% detection rate in a random sample of 100 samples.