Algorithm of stitching and fusion image in PCB analysis

Abstract

The article proposes an approach that allows merging a series of images obtained using an electron microscope and fusion data with a camera that records data in the far-infrared spectrum (thermal images). The proposed approach is implemented based on algorithms for stitching images obtained in the visible spectrum. Based on the data on the calibration parameters of the pair cameras, the images obtained by the thermal imaging camera are stitched. The frame overlap is between 30% and 50%. The data received by the thermal imaging camera is noisy and requires primary data processing. To filter thermal imaging images, a multi-criteria method is used in work. The parameters of the method are analyzed in parallel on both pairs of the image. They are used for the subsequent identification of the search boundaries of objects and interframe communication points. For data obtained in the visible spectrum, a simplification algorithm is applied to increase data analysis speed. As test data, we used a combination of images obtained by an electron microscope (a maximum approximation is 300x, the color depth is 8bit, the resolution is 1024x768) and a thermal imaging camera (image resolution is 320x240 pixels).