摘要
运动模糊图片的频谱中存在平行的零点直线。这里提出了一种算法,识别零点直线含有的运动模糊方向和模糊度信息:首先对模糊图像进行分块,然后将分成的小块相加,利用拉氏算子对最后的相加结果进行无方向性的二阶微分,通过二维FFT变换求得频谱。接着对频谱取绝对值、做归一化之后,修饰二值化结果,最后进行Radon变换,求得点扩展函数的运动方向和模糊长度。实验论证该算法可靠通用,性能优良。
There are parallel zero beelines in the spectrum image of motion-blurred image,A arithmetic put forward to identify the blurring direction and blurring length.First the motion-blurred image will be divided into small pieces,we sum all the pieces.After the second differentiation of the piece by Laplacian,it＇s Spectrum can be obtained by 2D-FFT transform,then modify the binary of spectrum.At the last time,we detect blurring direction and blurring length by Radon transform.The practicabi1ity and veracity of arithmetic has been proved by experiments.The proposed method has excellent capability.
出处
《微处理机》
2011年第1期59-63,共5页
Microprocessors
关键词
运动模糊
频谱
RADON变换
Motion blur
Frequency spectrum
Radon transform
