基于最小平均距离免疫算法的模糊红外图像分割

由于犯罪分子利用各种方法来避开传统的刑侦图像技术，因而红外图像逐渐成为获取犯罪现场痕迹的有效手段。然而，从犯罪现场拍摄的红外图像其目标痕迹大多是弱化的，所以在这类红外图像中分割目标是一项具有挑战性的任务。已有基于生物免疫的各类算法尚未明确描述免疫分割作用领域，以及免疫网络算法模型中的免疫识别距离。为实现对目标痕迹弱化红外图像的有效分割，提出了一种新的具有免疫作用领域和最小平均免疫识别距离的人工免疫构架，设计了一种具备最小平均距离免疫域的免疫分割算法。该方法根据红外图像的特点，采用多步分类算法、免疫变异和自适应免疫最小均距识别方法，根据目标区域和背景区域的总体统计特性实现最佳分类。实验结果表明，提出的基于最小平均距离的免疫算法能够有效地分割目标弱化的红外图像。与经典的边缘模板和区域模板方法相比，该算法具有更好的分割效果，尤其是针对目标弱化红外图像的分割，该算法能够较好地给出五个手指的边界轮廓。

Blurred Infrared Image Segmentation Using New Immune Algorithm with Minimum Mean Distance Immune Field

Criminals tend to use various methods to cope with the traditional forensic image technologies, so infrared image is becoming an effective means for obtaining crime scene traces. However, segmentation targets from infrared image shoot in crime scene is a challenging task as these images are target weakened infrared images. Previous studies about immune algorithms do not describe immune variation and immune recognition distance in the network and algorithm. In opposition to segment these target weakened traces infrared images, we propose a new immune framework with immune variation and minimum mean immune recognition distance, and construct a new immune segmentation algorithm with minimum mean distance immune field. According to the distinguishing feature of infrared images, this method use multi-step classification algorithm, immune variation and adaptive immune minimum mean distance recognition to achieve optimal classification based on the overall statistical properties of target areas and background areas. Experimental results show that the proposed immune algorithm with minimum mean distance can segment target weakened infrared images efficiently. Compared with classical edge template and conventional region template methods, the proposed algorithm has better segmentation results, especially the boundaries of five fingers.