像素
计算机视觉
计算机科学
图像(数学)
图像传感器
人工智能
计算机图形学(图像)
作者
Liu Yongji,Xiaojun Yuan
标识
DOI:10.1109/icaiis49377.2020.9194921
摘要
This paper introduces the problem of defective pixel in the field of image sensor and several methods of defective pixel detection and correction for the problem of dynamic defective pixel. It proposes a design of dynamic defective pixel correction, which divides the dynamic defective pixel correction into three steps: image edge expansion, defective pixel detection and defective pixel correction. Firstly, the edge of the image is expanded symmetrically. Then, the DIF and the AVG of 8 pixels in the same color neighborhood are calculated to detect the defective pixel. Finally, median correction is used to correct the defective pixel, with the calculation of direction gradient being used in the process of correction. The design involves simple calculation but has high detection rate of defective pixel. It also effectively improves the definition of the picture after correction.
科研通智能强力驱动
Strongly Powered by AbleSci AI