Recently, the reference functions for the synthesis and analysis of the autostereoscopic multiview and integral images in three-dimensional displays we introduced. In the current paper, we propose the wavelets to analyze such images. The wavelets are built on the reference functions as on the scaling functions of the wavelet analysis. The continuous wavelet transform was successfully applied to the testing wireframe binary objects. The restored locations correspond to the structure of the testing wireframe binary objects. © 2015 Optical Society of America The synthesis and analysis of images are two main concepts of the image processing. The real-time processing of the multiview images is probably most important problem of three-dimensional (3D) imaging and image processing in autostereoscopic multiview and integral displays [1], [2].The processing can be based on a set of functions. Such functions may be helpful for the real-time image synthesis and analysis. For instance, the computer-generated holograms can be synthesized from the hologram patterns [3], [4].The extraction of depth is one of particular problems of the analysis of multiview and integral images [5]. The depth can be extracted using various methods including the computational integral imaging reconstruction [6], [7].The autostereoscopic multiview images can be also synthesized from the voxel patterns [8]. The visual 3D images were confirmed, and the patterns were further developed as the reference functions [9]. The synthesis based on patterns is applicable for the real-time autostereoscopic three-dimensional displays.Using the mentioned reference functions, the structure of multiview images can be analyzed, and locations of voxels can be extracted. Basing on that, it is possible to build a depth map, which can be effectively used in the further processing of 3D multiview images, e.g., in 3D broadcasting.The above functions were applied for depth extraction under various conditions including noised and grayscale images and showed a good validity [10], [11]. However, the functions [9] are not orthogonal, they do not comprise a basis.A promising technique of the real-time processing is the wavelet analysis. Generally, the wavelets can be built by various methods. One of them is to start from the scaling functions [12]. The scaling functions should satisfy certain conditions, i.e., the two-scale relation, stability (Rietz function), and the partition of the unity. Then, a wavelet can be built as a linear combination of scaling functions [13] where ψ is a wavelet, φ is a scaling function, and β is a coefficient.A known scaling function which satisfies all necessary conditions is the Haar function [14]. The functions [9] are based on rectangular unit pulse which actually is the shifted Haar function.The Haar function itself defines a family of wavelets [15]. Similarly, the multiview reference functions can define wavelets. The preliminary considerations were presented at the conferences [16] and more. In the current paper, we develop such wavelets and...