按照二维函数的特点和视觉机制,提出了用来捕捉纹理基元的纹理检测器函数,基于纹理检测器和扩展的小波变换,提出了基于能量分解的影像纹理多尺度分析方法,并按照神经动力学的侧抑制和端点抑制等理论,实现了对多尺度纹理特征的融合，这一多尺度分析方法直接将影像纹理能量在时间一尺度空间分解,包含了相位信息,避免了基于线性变换多尺度分解引起的能量与相位分离,为纹理分析提供了一个层次性的框架,有效提高了纹理的识别能力。-According to the characteristics of two-dimensional functions and visual mechanisms proposed to capture the texture primitive texture detector function, based on texture detector and extended wavelet transform, a multi-scale energy decomposition analysis method based on image texture and follow nerve and endpoint lateral inhibition kinetics of inhibition theory, to achieve the integration of multi-scale texture features, this multi-scale analysis methods directly to the image texture energy in a time scale spatial resolution, including the phase information, to avoid the multi-linear transformation based on energy scale decomposition caused by the phase separation of texture analysis provides a hierarchical framework, improve the ability to identify the texture.