Depth Estimation for HDR Images

In this chapter, depth estimation for stereo pair of High Dynamic Range (HDR) images is proposed. The proposed algorithm consists of two major techniques namely conversion of HDR images to Low Dynamic Range (LDR) images or Standard Dynamic Range (SDR) images and estimating the depth from the converted LDR / SDR stereo images. Local based tone mapping technique is used for the conversion of the HDR images to SDR images. And the depth estimation is done based on the corner features of the stereo pair images and block matching algorithm. Computationally much less expensive cost functions Mean Square Error (MSE) or Mean Absolute Difference (MAD) can be used for block matching algorithms. The proposed algorithm is explained with illustrations and results.

PDE-Based Image Processing

This chapter describes the basic concepts of partial differential equations (PDEs) based image modelling and their applications to image restoration. The general basic concepts of partial differential equation (PDE)-based image modelling and processing techniques are discussed for image restoration problems. These techniques can also be used in the design and development of efficient tools for various image processing and vision related tasks such as restoration, enhancement, segmentation, registration, inpainting, shape from shading, 3D reconstruction of objects from multiple views, and many more. As a case study, the topic in consideration is oriented towards image restoration using PDEs formalism since image restoration is considered to be an important pre-processing task for 3D surface geometry, reconstruction, and many other applications. An image may be subjected to various types of noises during its acquisition leading to degraded quality of the image, and hence, the noise must be reduced. The noise may be additive or multiplicative in nature. Here, the PDE-based models for removal of both types of noises are discussed. As examples, some PDE-based schemes have been implemented and their comparative study with other existing techniques has also been presented.

3D Reconstruction of Graph Objects, Scenes, and Environments

The purpose of this chapter is to provide a basic understanding of how three-dimensional (3D) statistical visual displays aid in education. The chapter seeks to discuss the importance of surface objects, scenes, and environments reconstructed to enhance the interpretation of charts. Further described are the different types of 3D charts available: bar, line, and pie charts. The chapter also provides enlightenment about two new concepts: the “3D actual” and “3D obvious” charts. Overall, the chapter focuses on the theoretical background, pedagogical practice, usability, and applicability of using 3D surface charts. The chapter, in addition, provides explanations based on research done by Chikatla (2010), Dempsey, Chikatla, and Inpornvijit (2008), Fisher, Dempsey, and Marousky (1997), and Dempsey and Armstrong (1997).

Hybrid GPU Local Delaunay Triangulation through Points Consolidation

This chapter describes a surface reconstruction method that mixes interpolating and approximating features and its implementation in graphics hardware. Hybrid methods are useful in areas such sculpting, medicine, and cultural heritage, where details must be preserved. Such cases may also contain noise (due to sampling inaccuracies) or duplicated points (in the case of the scan is done from multiple points of view), where hybrid methods provide an interesting solution. The proposed method makes use of a point projection operator to create a regular distributed and noise free set of points, which is reconstructed using local Delaunay triangulations. Both points projection and triangulation methods are studied in its basic serial version, but aiming to design parallel versions (more specifically GPU implementations) that increase their performance. The adaptations required for the parallel reconstruction are discussed, and several implementation details are given.

Projective Geometry for 3D Modeling of Objects

This chapter surveys many fundamental aspects of projective geometry that have been used extensively in computer vision literature. In particular, it discusses the role of this branch of geometry in reconstructing basic entities (e.g., 3D points, 3D lines, and planes) in 3D space from multiple images. The chapter presents the notation of different elements. It investigates the geometrical relationships when one or two cameras are observing the scene creating single-view and two-view geometry. In other words, camera parameters in terms of locations and orientations, with respect to 3D space and with respect to other cameras, create relationships. This chapter discusses these relationships and expresses them mathematically. Finally, different approaches to deal with the existence of noise or inaccuracy in general are presented.

Methods of 3D Object Shape Acquisition

This chapter contains an overview of methods for a 3D object shape from both the surface and the internal structure of the objects. The acquisition methods of interest are optical methods based on objects surface image processing and CT/NMR sensors that explore the object volume structure. The chapter also describes some methods for 3D shape processing. The focus is on 3D surface shape acquisition methods based on multiple views, methods using single view video sequences, and methods that use a single view with a controlled light source. In addition, the volume methods represented by CT/NMR are covered as well. A set of algorithms suitable for the acquired 3D data processing and simplification are shown to demonstrate how the models data can be processed. Finally, the chapter discusses future directions and then draws conclusions.

3D Surface Reconstruction from Multiviews for Prosthetic Design

Existing methods that use a fringe projection technique for prosthetic designs produce good results for the trunk and lower limbs; however, the devices used for this purpose are expensive. This chapter investigates the use of an inexpensive passive method involving 3D surface reconstruction from video images taken at multiple views. The design and evaluation methodology, consisting of a number of techniques suitable for prosthetic design, is developed. The method that focuses on fitting the reference model (3D model) of an object to the target data (3D data) is presented. The 3D model is obtained by a computer program while the 3D data uses the shape-from-silhouette technique in an approximately circular motion. The modification of existing model-based reconstruction – mainly on the deformation process of vertices – is discussed, and the results of different objects show a good possibility for using a passive method in prosthetic devices. The methodology developed is shown to be useful for prosthetic designers as an alternative to manual impression during the design.

Application of Red, Green, and Blue Color Channels in 3D Shape Measurement

Optical full-field measurement techniques have been widely studied in academia and applied to many actual fields of automated inspection, reverse engineering, cosmetic surgery, and so on. With the advent of color CCD cameras and DMD (Digital Micromirror Device) based color DLP (Digital Light Processing) projectors, their major red, green, and blue channels have been used as a carrier to code fringe patterns. Since three fringe patterns can be simultaneously projected and captured at one shot, the acquisition time reduces to 1/3 of the value by the gray fringe pattern projection. This chapter will introduce two kinds of applications of red, green, and blue as a carrier: 1) modulation and demodulation method of coding sinusoidal fringe patterns into RGB channels of a composite color image; and 2) modulation and demodulation method of coding sinusoidal and binary fringe patterns into RGB channels of multiple composite color images. Experiments on testing the two kinds of applications were carried out by measuring the shape of objects’ surface. The results confirm that red, green, and blue channels can be used as a carrier to reduce the acquisition time.

Monocular-Cues Based 3-D Reconstruction

3-D reconstruction from images has traditionally focused on using multiple images. However, in recent years, some interesting breakthroughs have been made in constructing depth maps of images using monocular cues. This chapter summarizes the recent work in this evolving research domain. The chapter also presents results from an initial exploratory study based on fusing the 3-D reconstructions generated by two seminal monocular-cue based reconstruction algorithms. With new testing data, the fusion approach improved the 3-D estimation accuracy significantly as compared to the original approaches. The authors use the improved estimation accuracy produced by the fusion algorithm as a motivating evidence for future work: the use of non-parametric Bayesian regression for 3-D reconstruction.

Complementary Part Detection and Reassembly of 3D Fragments

This chapter presents an algorithm for identifying complementary site of objects broken into two parts. For a given 3D scanned image of broken objects, the algorithm identifies the rough sites of the broken object, transforms the object to a suitable alignment, registers it with its complementary part which belongs to the same object, and finds the local correspondence among the fragmented parts. The presented algorithm uses multiple granularity descriptors and edge extraction to detect the exact location of multiple cleavage sites in the object. It greatly reduces the amount of information to be matched and also helps in identification of the parts; as a result it reduces the computational time in the processing. It is also applicable to all triangulated surface data even in the presence of noise.