基于分级策略的自动人眼检测与定位

针对小波阈值法中的小波变换只能将图像分解到有限方向, 而不能较好地表征图像多方向性的问题, 用改进混合小波 方向滤波器组（HWD:Hybrid Wavelet Directional filter banks)变换代替单纯小波变换, 使在图像分解过程中更好地表征图像的多方向性, 保存更多的图像信息； 在分析小波阈值去噪原理的基础上,改变隶属度函数, 构建HWD隶属度的权系数, 从而避免因小波系数间存在幅值交叉使小波阈值法的应用受到限制。改进的HWD在损失最少图像小波系数的前提下, 能最大限度地置零噪声小波系数。实际工程图纸去噪研究表明, 改进的小波阈值法可在去除一定噪声的前提下, 保留更多的工程图纸细节信息。

Research on Drawings Denoising Based on Improved Wavelet Threshold Algorithm with Hybrid Wavelet-Directional Filter Transform

For the defects that the image can only be decomposed into a finite orientation by the wavelet transformation, and multi-directional of image can not be better characterized, the directional filter is applied to the method of wavelet threshold, namely, the HWD （ Hybrid Wavelet-Directional filter banks） instead of the mere wavelet transform is used to better characterized the multi-directional of image and to retain more image information. To avoid the phenomenon of amplitude cross exists between image wavelet coefficients and noise wavelet coefficients, the method of wavelet thresholding was restricted by this shortcoming, membership function was changed based on the principle of the wavelet, and the membership weights of HWD was built by the membership function. The wavelet coefficients of noise were set to zero maximum with losing the image coefficients at least by this improved HWD. The wavelet coefficients of noise were set to zero maximum with losing the image coefficients at least by this improved HWD. The studies show that more details of engineering drawings can be retained by the improved wavelet threshold method beside removing some noise.