摘要
在数字图像获取过程中由于机械振动、触碰等造成图像质量退化是非常普遍的现象。本研究将抖动模糊近似为匀速直线运动,对图像倒频谱取绝对值检测图像边缘,利用Radon变换的最大值求出模糊角度,并根据倒频谱图计算抖动模糊长度,数据显示运算误差在5个像素内。最后,通过参数容差,采用枚举法进行维纳滤波复原,并将最大熵值下的参数作为复原参数。实验仿真结果显示,图像复原效果比较理想,能够较好地提高图像质量。
Image quality fading is a very common phenomenon because of mechanical vibration or touching during digital im- age acquisition. In this study, the jitter fuzz was approximated for the uniform motion in a straight line, the absolute value of the cepstrum was used for edge detection, and the maximum of Radon transform was used to calculate the fuzzy angle. Then, by rotating the spectrum graph to calculate the fuzzy length, the data showed operation error is within 5 pixels. At last, the accurate parameters under the principle of maximum entropy value were calculated to be recovery parameter. The experimental simulation results showed that the effect of the image restoration is close to the ideal image.
出处
《中国印刷与包装研究》
CAS
2014年第5期37-43,共7页
China Printing Materials Market
基金
国家自然科学基金--基于多光谱的彩绘文物颜色褪变机理与虚拟复原(No.61405104)
曲阜师范大学本科生科研训练计划--基于多光谱的色彩管理方法研究(No.2012059)
