基于小波变换的深度预测复原算法研究

提出了一种基于小波变换的深度预测算法,并将此算法应用于计算光学切片的图像复原.对序列图进行一层小波分解,然后确定阈值和预测规则,对序列图进行预测.根据预测结果进行小波反变换,将图像复原.实验表明:小波系数应取三个高频分量的绝对值及部分低频信息之和;阈值Г随添加低频分量值的比例的增大而增大;采用微分算子规则具有较好的预测和复原效果.该方法计算量小、不需要迭代、不需要点扩展函数,针对没有交叠的序列图有较好的复原效果.

A Depth Estimation Restoration Algorithm Based on Wavelet Tranform

A depth estimation algorithm based on wavelet transform was proposed and applied in the image restoration of Computational Optical Sectioning Microscopy (COSM). Firstly, wavelet transform was performed on sequential images, and then the depth of each of images was estimated according to the threshold and estimation criterion. Finally, the sequential images were restored by inverse wavelet transform. The experiment presented that wavelet coefficients should consist of three high-frequency components as well as part of low-frequency component, the value of threshold Г was proportional to the low-frequency coefficients, and Difference Criterion was effective for better estimation and restoration effects. The advantage of this algorithm was its low computational cost, non-iteration and no knowledge of the Point Spread Function (PSF) of the imaging system. The algorithm had satisfying restoration effects for non-overlapped sequential images.