Research On Image Color Reconstruction Based On Laplacian And Sparse Representation

Image color reconstruction and enhancement,as two important research directions in the field of image processing,have attracted extensive attention from scholars.The automatic color reconstruction through grayscale images is a challenging research topic.In response to the above problems,this paper proposes the following algorithm:In the third chapter,proposes a new weight-based nonlocal Laplacian algorithm based on depth guidance for image color reconstruction.Compared with the traditional nonlocal weighted Laplacian method,the algorithm combines the luminance information and depth information of the image to propose a new weight function,and applies the nonlocal weighted Laplacian filter to each of the RGB images.A color channel that enhances color consistency with its depth similarity.In addition,the proposed method can be easily extended to the task of image depth enhancement.Under the guidance of color channel,the performance of the algorithm is significantly improved.A large number of experiments show that this method is superior to the traditional weighted nonlocal Laplacian algorithm.In the fourth chapter,proposes an instance-based color reconstruction algorithm based on local consistency sparse representation.The algorithm operates at the superpixel level,extracting low-level intensity features,intermediate texture features,and high-level semantic features for each superpixel,and joins them to form their descriptors.The set of feature vectors for all superpixels from the reference image are form a dictionary.In this paper,the coloring of target superpixels is understood as a dictionary-based sparse reconstruction problem,and a uniform regularization term is introduced in the energy formula to enhance locality and matching consistency.Finally,the brightness-guided joint filter is used to achieve smooth color propagation and maintain an accurate edge structure of the image,further enhancing color consistency.A large number of experimental results confirm the effectiveness of the proposed algorithm.