A point cloud is an effective 3D geometrical presentation of data paired with different attributes such as transparency, normal and color of each point. The imperfect acquisition process of a 3D point cloud usually generates a significant amount of noise. Hence, point cloud denoising has received a lot of attention. Most of the existing techniques perform point cloud denoising based only on the geometry information of the neighbouring points; there are very few works considering the problem of denoising of color attributes of a point cloud, and taking advantage of the correlation between geometry and color. In this article, we introduce a novel non-iterative set-up for the denoising of point cloud based on spectral graph wavelet transform (SGW) that jointly exploits geometry and color to perform denoising of geometry and color attributes in graph spectral domain. The designed framework is based on the construction of joint geometry and color graph that compacts the energy of smooth graph signals in the low-frequency bands. The noise is then removed from the spectral graph wavelet coefficients by applying data-driven adaptive soft-thresholding. Extensive simulation results show that the proposed denoising technique significantly outperforms state-of-the-art methods using both subjective and objective quality metrics.

Accurate face segmentation strongly benefits the human face image analysis problem. In this paper we propose a unified framework for face image analysis through end-to-end semantic face segmentation. The proposed framework contains a set of stack components for face understanding, which includes head pose estimation, age classification, and gender recognition. A manually labeled face data-set is used for training the Conditional Random Fields (CRFs) based segmentation model. A multi-class face segmentation framework developed through CRFs segments a facial image into six parts. The probabilistic classification strategy is used, and probability maps are generated for each class. The probability maps are used as features descriptors and a Random Decision Forest (RDF) classifier is modeled for each task (head pose, age, and gender). We assess the performance of the proposed framework on several data-sets and report better results as compared to the previously reported results.

Artificial intelligence (AI) is the main branch of computer science that permits advanced machines to interpret and analyze complex healthcare data elaborating the recent challenges in the medical field of study. The current state of AI applications in healthcare is examined in this systematic literature review, with an emphasis on the technology's accomplishments, difficulties, and potential. The wide breadth of AI technologies used in healthcare settings, such as robots, computer vision, machine learning, and natural language processing, is highlighted in this review through an extensive analysis of peer-reviewed publications. It talks about how customized medicine, predictive analytics, illness detection, and treatment planning are just a few of the areas of healthcare delivery that AI-driven technologies are transforming. According to research by investment bank Goldman Sachs, 300 million full-time employees could be replaced by artificial intelligence (AI). In the US and Europe, it might replace 25% of labor duties, but it might also lead to an increase in productivity and the creation of new jobs. Additionally, it might eventually result in a 7% rise in the global annual value of products and services produced. Additionally, the paper projects that approximately 25% of all employment might be performed totally by AI and that two-thirds of jobs in the U.S. and Europe "are exposed to some degree of AI automation. "The most likely groups to be impacted by workforce automation are educated white-collar workers making up to $80,000 annually, according to research from OpenAI and the University of Pennsylvania. According to a McKinsey Global Institute study, developments in digitalization, robots, and artificial intelligence may require at least 14% of workers worldwide to change jobs by 2030.

This work presents a double asymmetric data hiding technique. The first asymmetry is created by hiding secret data in the complex region of the cover image and keep the smooth region unaffected. Then another asymmetry is developed by hiding a different number of secret bits in the various pixels of the complex region. The proposed technique uses the ant colony optimization (ACO) based technique for the classification of complex and smooth region pixels. Then, the variable least significant bits (VLSB) data hiding framework is used to hide secret bits in the complex region of the cover image. A distance-based substitution technique, namely increasing distance increasing bits substitution algorithm, is used to ensure the asymmetry in the number of hidden bits. The double asymmetric framework enhances the security of the hidden secret data and makes the retrieval of hidden information difficult for unauthorized users. The algorithm results in high-quality stego images, and the hidden information does not attract the human visual system (HVS).

Accurate volumetric segmentation of 3D medical imaging modalities is critical for therapy planning and clinical diagnosis, particularly for brain tumor delineation. Traditional convolutional neural network (CNN)-based architectures face challenges while capturing global contextual information and modeling long-range dependencies in complex 3D volumetric data, limiting their segmentation performance. Transformer-based models have emerged as promising alternatives to CNNs for such tasks, addressing their limitations in capturing global spatial dependencies. We propose 3D-ViT-UNet, a novel U-shaped vision transformer (ViT)-based encoder-decoder architecture for end-to-end volumetric brain tumor segmentation. The model employs 3D Window Multi-Head Self-Attention (3D-W-MSA) to capture local features and a 3D Dilated-Window Multi-Head Self-Attention (3D-DW-MSA) to capture global features while reducing computational complexity. Moreover, for preserving absolute and relative positional information and preventing permutation equivalence limitation in transformers, a dynamic position encoding strategy is integrated. The proposed model demonstrates state-of-the-art (SOTA) performance for brain tumor segmentation on the BraTS 2020 dataset. It achieves a superior average Dice Similarity Coefficient (DSC) of 84.81% and a Hausdorff Distance (HD) of 4.87 mm with reduced computational complexity compared to existing methods. Also, an improvement in delineation of tumor boundaries and accurate segmentation across modalities is demonstrated through the qualitative results. Extensive quantitative and qualitative evaluations highlight the capability of 3D-ViT-UNet to achieve high accuracy with a smaller model size and lower FLOPs, making it an effective and efficient solution for clinical applications involving volumetric brain tumor segmentation.