利用红外热像技术对PCB（Printed Circuit Board，印制电路板）的板载元件工作状态进行探测，是一种新型的非接触式的检测方法。针对传统检测方法未将PCB 的可见光图像作为检测信息的有益补充，本文把两种图像进行融合，将有用的信息有机地结合了起来。首先检测可见光与红外图像的最大稳定极值区域特征，利用简化的SURF 描述子实现多模态图像的配准。接下来采用结合非下采样轮廓波变换和压缩感知技术进行红外与可见光图像融合，获得兼具元器件位置和工作属性等信息的图像。最后使用ID3 决策树获得被测故障的最大的类别信息，通过实验表明了基于信息融合的红外热像技术进行PCB 检测的可行性，同时也对局限性进行了分析和阐述。

A new type of non-contact detection method has emerged that uses infrared thermal imaging technology to detect the working state of printed circuit board (PCB) components. In traditional detection methods, a PCB's visible image is not used to supplement detection information. In this study, multi-modal information is organically combined. First, the maximum stable extremum regions of visible and infrared images are extracted, and then simplified speeded-up robust features are descripted for multi-modal image registration. Then, the combined nonsubsampled contourlet transform and compressed sensing technology are used for image fusion to obtain images with the component location and performance status. Finally, the ID3 decision tree is used to split the classification of the fault attributes. An experiment conducted in our study shows the feasibility of PCB detection based on infrared thermal imaging. In addition, the limitations are analyzed and expatiated.