宽幅扫描仪色差一致性校正与边缘融合拼接

Fusion and Combination of Margins for Large Format Scanner

摘要: 针对CIS (Contact Image Sensor)宽幅扫描仪多根CIS传感器之间存在的物理差异导致采集到图像存在色彩差异的问题，提出了一种基于多传感器像素映射的图像色差一致性校正算法。针对特殊的工程图纸存在底色干扰问题，提出了一种底色消除像素映射算法，以最大可能的保留图像内容和细节。由于多根CIS传感器安装位置存在交叉，导致直接采集到的图像边缘有像素重叠现象，需要对采集图像进行融合与拼接。本文提出了一种行匹配的窗口算法实现图像的融合拼接。测试结果表明，色差一致性校正算法、底色消除算法和图像边缘融合拼接算法能有效解决CIS宽幅扫描仪技术问题，具备良好的算法性能与图像处理效果。

Abstract: A correction algorithm of chromatic aberration is proposed by mapping functions for different  sensors, which have physical differences leading to chromatic aberration, when scanned by CIS. A new algorithm of pixel mapping function is proposed, which can eliminate the background noise of drawings and retain some important contents and details. An algorithm of matching line window is also proposed, which can solve the problem of image fusion and combination. It is called pixel overlap from two images due to intersecting location of sensors. The final test shows that the algorithm of chromatic aberration and the algorithm of pixel mapping function can solve the key problems by CIS effectively with good performance and image processing effects.