In criminal investigation, there are hidden traces that many people can’t find, so infrared image is becoming an effective means to obtain these scene traces. The extraction algorithm with growth immune field can extract the target of infrared image relatively effectively, but it is lack of efficiency and reliability in complex environment. Here we propose a new target extraction algorithm with adaptive growth immune field, combining the image information of region and edge gradient. The region of the target in complex environment is obtained by K-means clustering algorithm and the source seed points are selected from the region. The regional characteristics around the seed points as the criteria for growth and the image gradient information is applied as the condition of the adaptive growth immune field. This algorithm improves the accuracy of target extraction in complex environment while preventing overgrowth. We compare the algorithm with the original algorithm and other algorithms and we find that the new algorithm combining edge gradient information can reduce the probability of over growth and ensure the integrity of target extraction under complex background.

中文翻译：

---

基于自适应生长免疫场的红外图像提取算法

在刑事调查中，存在许多人找不到的隐藏痕迹，因此红外图像正成为获取这些场景痕迹的有效手段。具有生长免疫场的提取算法可以相对有效地提取红外图像的目标，但是在复杂环境下却缺乏效率和可靠性。这里我们提出了一种新的具有自适应增长免疫场的目标提取算法，将区域和边缘梯度的图像信息相结合。通过K-means聚类算法获得复杂环境中目标的区域，并从该区域中选择源种子点。将种子点周围的区域特征作为生长的标准，并将图像梯度信息用作适应性生长免疫场的条件。该算法提高了复杂环境中目标提取的精度，同时防止了过度增长。将算法与原始算法和其他算法进行比较，发现结合边缘梯度信息的新算法可以降低过增长的可能性，并能确保复杂背景下目标提取的完整性。