一种基于红外图像的目标自动识别算法

A Method of Automated Recognition and Classification based on Infrared Images

红外成像制导具有在各种复杂战术环境下自主搜索、捕获、识别和跟踪目标的能力,代表了当代红外制导技术的发展趋势。提出了一种红外图像预处理、跟踪、分类的自动目标识别算法,利用小波变换、形态学方法对红外图像进行预处理,提取不同频带的惯性不变矩作为特征量,利用神经网络进行分类识别,结果表明该算法具有很高的识别率,对于精确制导武器的目标识别研究具有一定的参考价值。

Infrared imaging homing is development trend of infrared homing technology, which has independent capability in target recognition and tracking. A novel ATR algorithm including image pre - processing, tracking and classification for sequence of infrared （IR） targets is presented in this paper. Decompose the infrared image by wavelet transform, and then extract the moments as the features to be recognized by BP neural network. The experimental result indicates that this method has very high discernment rate and definite practical value to target recognition study of accurate guidance weapon.