基于圆对称性的离焦模糊图像还原方法

针对离焦模糊图像,圆盘离焦模型和高斯离焦模型的还原效果并不理想,提出基于圆对称性的离焦模糊图像还原算法。通过刃边函数估算模糊半径,利用信噪比还原图像,进而计算图像的信号功率谱。提出改进的DCT分块加权清晰度评价函数,对DCT系数作指数加权优化,并根据圆对称性模型检测出的多个模糊半径还原出多个图像,并筛选出最清晰图像作为还原结果。实验结果显示,与传统的还原方法相比,该方法提高了还原图像的清晰度和分辨率。

Restoration of Defocus Blurred Image Based on Circular Symmetry

For the defocus blurred image, the restoration effect of disc defocus model and Gaussian defocus model is not satisfied. This paper presents defocus blur image restoration algorithms based on circular symmetry. It estimats blur radius by blade edge function, restore the image by using SNR, and then calculate the signal power spectrum of the image. This paper presents the improved weighted DCT block definition evaluation function. The DCT coefficients are optimized by exponential weighted. The clearest image is chosen as result by using several blur radius detected from circular symmetry model. The experimental results show that, compared with the traditional restoration methods, the proposed method improves the image clarity and resolution.