Region texture analysis based Laplacian pyramid color transfer ICA method

Independent Component Analysis(ICA) applied in the field of image processing is a noval transformation domain method that utilizes sparse coding on the basis of analyzing the characteristics of the human visual system. It has multidirectionality, feature extraction, and edge modeling characteristics. Color transfer is currently the best way to integrate natural color in images. On the basis of effectively integrating ICA method and color transfer algorithm, a region texture color transfer-ICA natural color fusion algorithm is proposed by organically combining the matching parameters and transfer parameters of color transfer. Dynamic online method training ICA domain decomposition kernel function and synthesis kernel function; generating grayscale fusion images according to regional energy fusion rules; using grayscale image color transfer algorithm based on regional Gray Level Co-occurrence Matrix(GLCM) texture to extract texture features of reference images, achieving optimal matching with regional texture features of grayscale fusion images, and linearly assigning first-order and second-order color information to grayscale images to generate source color image; the Laplacian pyramid decomposes the color space channels of the source color image and color reference image into multiple resolutions for color transfer, enhancing the natural color fusion image’s representation of significant scene information such as local textures. Human visual perception and objective evaluation indicates that the fusion image highlights band features and enhances detailed information with natural and comfortable colors, further improving scene perception.