Abstract Precise three‐dimensional (3D) instance segmentation of indoor scenes plays a critical role in civil engineering, including reverse size detection, and advanced structural analysis. However, existing methods often fall short accurately segmenting complex environments due to challenges diverse material textures, irregular object shapes, inadequate datasets. To address these limitations, this paper introduces StructNet3D, point cloud neural network specifically designed for components...

In this paper, we present local patterns constrained image histograms (LPCIH) for efficient retrieval. Extracting information through combining texture with global histogram, LPCIH is an effective feature representation method a flexible segmentation process. This kind of robust and invariant several transforms, such as rotation, scaling damaging. difficult retrieval tasks, rotated damaged gray Because many traditional methods are not suitable the valuable real-world applications...

Illumination is an important factor to affect image quality in computer vision. In order effectively weaken the influence of illumination on quality, correction algorithm based tone mapping proposed this paper. The combined color space decomposition and brightness adjustment, which can improve contrast while maintaining better original image, cannot increase noise. On other hand, pixel processing does not need transformation statistics information, reduce computational complexity more easily...

In this article, we present a ball detection algorithm based on color information and Hough transform. Our objective is to detect the balls under complex environments different illumination conditions. To do so, firstly apply multiscale Retinex enhance image. Then, an adaptive transform introduced locate position of in order reduce influence environments. Thirdly, interested conditions, introduce invariant model Grayworld normalization (GN) overcome illumination. Finally, evaluate proposed...

Recently, digital image inpainting technique based on variational and partial differential equation has been researched widely. In this paper, we present two compression schemes technologies. One scheme in spatial domain is the edge character. The Harmonic applied scheme. other wavelet domain. scheme, p-Laplace operator used inpainting. these schemes, mainly focus decoding step using technology. Experimental results showthe feasibility of schemes. Compared with JPEG2000, our could save a few...

It has been widely acknowledged that learning-based super-resolution (SR) methods are effective to recover a high resolution (HR) image from single low (LR) input image. However, there exist two main challenges in SR currently: the quality of training samples and demand for computation. We proposed novel framework tasks aiming at these issues, which consists blind blurring kernel estimation (BKE) recovery with anchored space mapping (ASM). BKE is realized via minimizing cross-scale...

Visual quality measure is one of the fundamental and important issues to numerous applications image video processing. In this paper, based on assumption that human visual system sensitive structures (edges) local luminance (light stimulation), we propose a new perceptual assessment (PIQA) total variation (TV) model (TVPIQA) in spatial domain. The proposed compares TVs between distorted its reference represent loss structural information. Because good performance TV describing edges, TVPIQA...

In this paper, based on the modified Hopfield neural network, we present two variational PDEs (Partial Differential Equations) as regularization terms to image restoration model. One is a harmonic model and other total variation Then, propose novel algorithms Modified Neural Network (MHNN). Both are aiming restore degraded images preserve edges with improved visual quality. The experimental results demonstrate that our proposed perform better than known network algorithms.

<abstract> The purpose of this paper is to investigate the existence weak solutions for a Kirchhoff-type problem driven by non-local integro-differential operator as follows: <p class="disp_formula">$ \begin{equation*} \left\{ \begin{aligned} &amp;M\left(\int_{\mathbb{R}^{2N}}\frac{|u(x)-u(y)|^{p(x,y)}}{p(x,y)|x-y|^{N+sp(x,y)}}dxdy\right)(-\Delta_{p(x)})^{s}u(x) = f(x,u)&amp;{\rm{in}}\,\,&amp;\Omega,\\ &amp;\quad u 0 &amp;{\rm{in}}\,\,&amp;\mathbb{R}^{N}\backslash\Omega,\\...

The video enhancement is necessary and needed in computer vision many applications, especially low illumination environment. In this paper, a fast method based on simplified VEC (Virtual Exposure Camera) model proposed to enhance dim videos. mainly includes two parts. One part the filters that we used pre post processing reduce noise. other tone mapping adjust illumination. Experimental results show effective real time processing, satisfied for exposure video.

Most image inpainting techniques require a human to identify manually the portion be inpainted. We propose new detection algorithm which can automatically detect defect portions in an image, including arbitrary scratches by color ink. The proposed is based on characteristics of uniform gray and structural information scratches. In this algorithm, we apply top-hat transform low-hap transformation highlight scratch regions use erosion- dilation operator morphology finalize Many photos are used...

In our previous work, an image restoration algorithm based on modified Hopfield neural network and harmonic model was proposed. But the computational complexity of is high. this paper, a parallel Harmonic using (MHNN) developed. The proposed makes full use computation performance effectively reduces from O(n×L) to O(n). Numerous experimental results demonstrate that outperforms traditional in terms complexity.

Multi-touch is a novel and vigorous Human Computer Interaction technique, which has being widely applied in both education industry. This article describes the implementation of Mimosa — standing multi-touch system that primarily designed for exhibiting products or displaying virtual scenes public space conditions. based on computer vision its whole steps are discussed detail paper. We firstly propose hardware setup suitable environments, further analyze software architecture bridging gap...