Abstract: Aiming at the defects existing in wavelet threshold denoising that it is difficult to select threshold and the quantization effect of threshold function is bad, an image denoising method based on artificial bee colony algorithm and parametric threshold function is proposed. Firstly, a new wavelet threshold function is designed, which has continuity, high-order differentiability and parameter adjustability. It can effectively solve the problems of the discontinuity in hard threshold function and the constant deviation in soft threshold function. Secondly, this method selects the optimal threshold through the artificial bee colony optimization algorithm, and then substitutes it into the new wavelet threshold function to denoise the noisy image. Finally, to compare the image denoising effects between the new threshold function and the traditional threshold functions, simulation experiment with MATLAB is carried out. The experimental result shows that the proposed threshold function based on ABC algorithm has a better image denoising effect than the traditional ones.

Key words: image denoising, wavelet transform, threshold function, Artificial Bee Colony（ABC） algorithm

摘要： 针对小波阈值去噪方法中存在阈值选取困难和阈值函数量化效果差的缺陷，提出一种基于人工蜂群算法和带参阈值函数的图像去噪方法。首先，设计一个新的小波阈值函数，该函数具有连续性，高阶可微性和参数可调性，能够有效地解决硬阈值函数的不连续性和软阈值函数具有恒定偏差的问题。然后采用人工蜂群优化算法选取最优阈值，将其代入新小波阈值函数对带噪图像进行去噪处理。最后用MATLAB进行仿真实验，对比新阈值函数和传统阈值函数的去噪效果。实验结果表明：在图像去噪效果方面，提出的基于人工蜂群算法的新阈值函数明显优于传统阈值函数。

关键词: 图像去噪, 小波变换, 阈值函数, 人工蜂群算法