Project topics and materials
Get Complete Project Material File(s) Now! »
Basics of Image Compression Techniques
Digital image and video compression is the science of coding the image content to reduce the number of bits required in representing it, aiming facilitate the storage or transmission of images with a level of quality required for given application (digital cinema, mobile video streaming) [19]. Typically, a digital image signal contains visual information in a two-dimensional matrix of size equal to N rows by M columns. Each spatial sample also known as a pixel is represented digitally with a finite number of bits called bit-depth [20].
For example, each pixel in a grayscale image is typically represented by a byte word, i.e. 8 bit-depth. A standard RGB color image is represented by three byte words, i.e. 24 bit-depth corresponding to 8 bits for the red component, 8 bits for the green one, and 8 bits for the blue one [21]. Digital video signals are represented as a collection of successive still images separated by a fixed interval time, which determines the so-called frame rate [16].
The compression process can be performed by exploited many duplicated information in the digital image or video signals. For example, it is possible to exploit the fact that the human eye is more sensitive to brightness than color for reducing the size of an image. To do that, the RGB components of the color image are first converted into the three YUV color components, where Y corresponds to the luminance (brightness) and U and V are the chrominance components, respectively. Then, the chrominance components are usually reduced by a factor of 1,5 or 2 by appropriate spatial down sampling [22].
Illustrative Example of the JPEG Still Image Compression Standard
The JPEG compression standard is one of the most well-known image compression standards. It takes its name from the working group called the Joint Photographic Expert Group that developed it in the early 1990s. Today, the JPEG standard is still widely used [26] in a broad range of digital imaging applications like digital photography, medical imaging, or video recording (using Motion JPEG) [26]. Moreover, it provides the basis for future standards including JPEG2000, and High Efficiency Video Coding (HEVC)-Intra. JPEG is designed to handle color and grayscale image compression with an achieved compression ratio of up to 1: 100 [36] . It is based on the Discrete Cosine Transform (DCT) which analyses the image as the human eye does. The human eye does not see all the colored details present in the image, consequently the fine details corresponding to high spatial frequencies can be removed with no effect for the human viewer [37]. The encoding process is started by dividing the original image into squared blocks of 8×8 samples. Each block is transformed by Forward DCT or DCT from the pixel domain to the frequency domain in order to reduce the spatial redundancy. After DCT, the block energy is generally concentrated in few low frequency transform coefficients. Then, the 64 DCT coefficients are quantized hence reducing the number of non-null values. Finally, the quantized coefficients are sent to the entropy coder that delivers the output stream of compressed image data as illustrated in Figure 2-4.
Motion Compensation and Video Compression
Digital video compression is the process that aims to reduce the spatiotemporal redundancy contained in successive video frames to achieve a given bit rate [41]. The primary constraints concern the quality of the decoded video must satisfy specific requirements and the computational complexity involved in the operation. In order to exploit temporal redundancy, a video coder incorporates an additional Motion Estimation (ME)/motion compensation (MC) process. ME aims at estimating the displacement parameters of moving objects between two consecutive frames, while MC exploits these parameters to match the objects along the temporal axis. ME/MC has proven its efficiency in digital video processing and has become the core component of digital video compression technologies such as MPEG, H.264/AVC, and HEVC for removing the temporal redundancy. The concept of motion-compensated codec presents in following classic codec scheme in Figure 2-5.
Table of contents :
LIST OF FIGURES
LIST OF TABLES
CHAPTER 1 INTRODUCTION
1.1 Context and Motivation
1.2 Challenges
1.3 Contributions
1.4 Structure of the Manuscript
CHAPTER 2 STATE-OF-THE-ART
2.1 Digital Images and Video Compression
2.1.1 Basics of Image Compression Techniques
2.1.2 Illustrative Example of the JPEG Still Image Compression Standard
2.1.3 Motion Compensation and Video Compression
2.2 Material Surface Engineering
2.2.1 Surface Topography
2.2.2 Surface Topography Measurement
2.2.3 Mechanical Image Deformation Analysis
2.2.4 Surface Topographical Images Classification
2.3 Support Vector Machine (SVM)
2.3.1 Mathematical Linear SVM
2.3.2 Nonlinear SVM
2.3.3 K-Fold Cross-Validation
2.3.4 Multiclass SVM
2.3.4.1 One-Against-All (OAA)
2.3.4.2 One-Against-One (OAO)
2.4 Conclusion
CHAPTER 3 HIGH EFFICIENCY VIDEO CODING (HEVC)
3.1 Improvements in HEVC Coding Stages
3.2 HEVC Intra Prediction Coding
3.3 Lossless Coding
3.4 High Bit Depth Still Picture Coding
3.5 Conclusion
CHAPTER 4 PERFORMANCE EVALUATION OF STRAIN FIELD MEASUREMENT BY DIGITAL IMAGE CORRELATION USING HEVC COMPRESSED ULTRA-HIGH-SPEED VIDEO SEQUENCES.
4.1 Context of the study
4.2 Methodology
4.3 Methods and Materials
4.3.1 High-speed test device
4.3.2 HEVC Lossy and Lossless Compression
4.4 Results
4.4.1 Tensile Test of Polypropylene (PP) Specimen
4.4.2 Sikapower Arcan test
4.4.3 Discussion
4.5 Conclusion
CHAPTER 5 SVM CLASSIFICATION OF MULTI-SCALE TOPOGRAPHICAL MATERIALS IMAGES IN THE HEVC-COMPRESSED DOMAIN
5.1 Context of the study
5.2 Methodology
5.2.1 Methods and Materials
5.2.2 Surface Processing
5.2.3 Topographical Materials Texture Image Dataset
5.2.4 IPHM-Based Classification
5.2.5 HEVC Lossless 4×4 PU Compression
5.2.6 SVM Classification
5.3 Results
5.3.1 The Impact of Surface Topography Filtering Types on Achieved Compression Ratios
5.3.2 Evaluating IPMH As Texture Feature Descriptor
5.3.3 The Impact of Surface Topography Filtering Types on Topographical Images Classification Accuracy
5.3.4 The Impact of Scale of Analysis on Topographical Images Classification Accuracy
5.4 Conclusion
CHAPTER 6 CONCLUSION AND PERSPECTIVES
REFERENCES .
