Directional Subimages Research Articles

Waypoint Transfer Module between Autonomous Driving Maps Based on LiDAR Directional Sub-Images

Lane graphs are very important for describing road semantics and enabling safe autonomous maneuvers using the localization and path-planning modules. These graphs are considered long-life details because of the rare changes occurring in road structures. On the other hand, the global position of the corresponding topological maps might be changed due to the necessity of updating or extending the maps using different positioning systems such as GNSS/INS-RTK (GIR), Dead-Reckoning (DR), or SLAM technologies. Therefore, the lane graphs should be transferred between maps accurately to describe the same semantics of lanes and landmarks. This paper proposes a unique transfer framework in the image domain based on the LiDAR intensity road surfaces, considering the challenging requirements of its implementation in critical road structures. The road surfaces in a target map are decomposed into directional sub-images with X, Y, and Yaw IDs in the global coordinate system. The XY IDs are used to detect the common areas with a reference map, whereas the Yaw IDs are utilized to reconstruct the vehicle trajectory in the reference map and determine the associated lane graphs. The directional sub-images are then matched to the reference sub-images, and the graphs are safely transferred accordingly. The experimental results have verified the robustness and reliability of the proposed framework to transfer lane graphs safely and accurately between maps, regardless of the complexity of road structures, driving scenarios, map generation methods, and map global accuracies.

Statistical model and local binary pattern based texture feature extraction in dual-tree complex wavelet transform domain

This paper presents a new feature extraction method in dual-tree complex wavelet transform domain. Given an input image, we obtain all highpass directional subimages and a set of pyramid lowpass subimages with different resolutions by applying DTCWT decomposition. After that, generalized Gamma density $$(\hbox {G}\Gamma \hbox {D})$$ models and local binary pattern are utilized respectively to characterize features of both highpass and lowpass subimages. The two kinds of features are combined for texture classification, and the experimental results on datasets Brodatz, Outex and UMD demonstrate that our proposed method can achieve superior classification accuracy than other state-of-the-art methods.

Zero Watermarking Method Based on Enhanced Bit-Plane

Traditional watermark embedding introduces inevitably some perceptible quality degradation of the host image. Another problem is the inherent conflict between imperceptibility and robustness. However, the zero-watermarking technique can extract some essential characteristics from the host image and use them for watermark registration and detection. The original image was decomposed into series of multiscale and directional subimages after lifting wavelet transformation (LWT). The high order bit-plane of low-frequency subimage and watermark image are inputs of the cellular neural network (CNN), and the zero-watermarking registration image is the output. To investigate and improve the security and robustness, the original watermark and registration image are scrambled or encrypted. The watermark image can be extracted from the secret image. This algorithm is simple and robust. The proposed method is also simple for hardware realization.

A blind watermarking algorithm for fingerprint images based on contourlet transform

We present a novel blind biometric watermarking algorithm based on contourlet transform in this paper. The host image is decomposed into a series of multiscale, local and directional subimages using contourlet transform. A high frequency directional sub-band which has the highest energy is used for watermarking. At the same time, the original binary watermark is scrambled by two-dimensional Arnold transform, and then embedded into the selected directional sub-band. The signs of the contourlet coefficients are modified depending on the watermark bit. The watermark retrieving algorithm is a blind detection process, and it does not need the original image. Iris code is taken as the watermark and finger print image is taken as the host image. The simulation results show that the proposed watermarking algorithm gives good perceptual transparency and is able to resist additive channel noise attacks, cropping, histogram equalisation, gamma correction, etc.

Image zero-watermarking algorithm based on Contourlet transform and singular value decomposition

An image zero-watermarking algorithm was proposed based on contourlet transform and singular value decompositon. After the coutourlet transform, the original image was decomposed into series of multi scale, local and directional subimages. The singular value of low frequency component is decomposed. The watermark is embedded and extracted from the first singular value of the every block of image. Experimental results show that the digital image watermarking Algorithm has a good transparence of embedded watermark image and is robust against attacks such as added noise, JPEG compression and cropping. The reliability of watermark extraction was enhanced.

WAVELET DIRECTIONAL HISTOGRAMS OF THE SPATIO-TEMPORAL TEMPLATES OF HUMAN GESTURES

This paper evaluates the efficacy of directional information of wavelet multi-resolution decomposition to enhance histogram-based classification of human gestures. The gestures are represented by spatio-temporal templates. This template collapses spatial and temporal components of motion into a static gray scale image such that no explicit sequence matching or temporal analysis is required, and it reduces the dimensionality to represent motion. These templates are modified to be invariant to translation and scale. Two-dimensional, 3-level dyadic wavelet transforms have been applied on the template resulting in one lowpass sub-image and nine highpass directional sub-images. Histograms of wavelet coefficients at different scales are used for classification purposes. The experiments demonstrate that while the statistical properties of the template provide high level of classification accuracy, the global detail activity available in highpass decompositions significantly improve the classification accuracy.

Identifying the characteristic scale of scene variation in fine spatial resolution imagery with wavelet transform-based sub-image statistics

Understanding the spatial structure of fine spatial resolution images is instrumental for either pixel- or object-based image analysis. In this Letter, the characteristic scale of scene variation in images is evaluated using statistics of sub-images produced by a wavelet transform. Six statistics, namely mean, variance, standard deviation (SD), coefficient of variation, skewness and kurtosis, were calculated for three directional sub-images and their derivative energy signature image at a sequence of wavelet decomposition levels. A simulated image, an aerial AUSIMAGE™ image (spatial resolution 0.2 m) and a recent Système Probatoire de l'Observation de la Terre (SPOT)-5 High Resolution Geometric (HRG) panchromatic image (spatial resolution 2.5 m) were analysed. It was found that with energy signature images, the change rate of SD over spatial resolution ranges between two successive decomposition levels ( ( SD/ ( R) suggested a synoptic and approximate description for the characteristic scale of scene variation. However, by comparing the result with the ranges of geostatistical variograms, it is suggested that the geostatistical method can correctly identify the characteristic scales of scene variation; semivariances can be calculated at any lag and orientation, while standard wavelet transforms are decomposed at only limited spatial resolution and orientation levels.

Multiresolution motion estimation techniques for video compression

The wavelet transform is a valuable tool in video processing because of its flexibility in representing nonstationary signals. Waveletbased compression has the advantages of efficient decorrelation of image frames and reduced-complexity multiresolution motion estimation (MRME). We propose three techniques to improve motion estimation in a wavelet-based coder. First, we propose to use an adaptive threshold for coding the motion vectors of the high-pass subimages. Secondly, we propose a bidirectional motion estimation (BMRME) technique in the wavelet transform domain. In BMRME, we estimate the temporal (i.e., direction information) flags only for the blocks in the lowest-resolution subimages and use the same information for the corresponding blocks in the higher-resolution subimages. Finally, we propose a fast multiresolution motion estimation technique where the directional subimages at each level of the wavelet pyramid are combined into a single subimage. Multiresolution motion estimation is then performed on the newly formed subimages. The proposed techniques improve the coding performance significantly over the baseline MRME technique. In addition, they further reduce the computational complexity of the MRME technique.

A new stereo vision model based upon the binocular fusion concept

In computer vision, the idea of using stereo cameras for depth perception has been motivated by the fact that in human vision one percept can arise from two retinal images as a result of the process called “fusion”. Nevertheless, most of the stereo algorithms are generally concerned with finding a solution to obtaining depth and three-dimensional shape irrespective of its relevance to the human system. Recent progress in the study of the brain mechanisms of vision has opened new vistas in computer vision research. This paper investigates this knowledge base and its applicability to improving the technique of computer stereo vision. In this regard, (1) a stereo vision model in conjunction with evidences from neurophysiology of the human binocular system is established herein; (2) a computationally efficient algorithm to implement this model is developed. This algorithm has been tested on both computer generated and real scene images. The results from all directional subimages are combined to obtain a complete description of the target surface from disparity measurements.