Patch‐based Texture Interpolation

In this paper, we present a novel exemplar‐based technique for the interpolation between two textures that combines patch‐based and statistical approaches. Motivated by the notion of texture as a largely local phenomenon, we warp and blend small image neighborhoods prior to patch‐based texture synthesis. In addition, interpolating and enforcing characteristic image statistics faithfully handles high frequency detail. We are able to create both intermediate textures as well as continuous transitions. In contrast to previous techniques computing a global morphing transformation on the entire input exemplar images, our localized and patch‐based approach allows us to successfully interpolate between textures with considerable differences in feature topology for which no smooth global warping field exists.     

Peek‐in‐the‐Pic: Flying Through Architectural Scenes From a Single Image*

Many casually taken ‘tourist’ photographs comprise of architectural objects like houses, buildings, etc. Reconstructing such 3D scenes captured in a single photograph is a very challenging problem. We propose a novel approach to reconstruct such architectural scenes with minimal and simple user interaction, with the goal of providing 3D navigational capability to an image rather than acquiring accurate geometric detail. Our system, Peek‐in‐the‐Pic, is based on a sketch‐based geometry reconstruction paradigm. Given an image, the user simply traces out objects from it. Our system regards these as perspective line drawings, automatically completes them and reconstructs geometry from them. We make basic assumptions about the structure of traced objects and provide simple gestures for placing additional constraints. We also provide a simple sketching tool to progressively complete parts of the reconstructed buildings that are not visible in the image and cannot be automatically completed. Finally, we fill holes created in the original image when reconstructed buildings are removed from it, by automatic texture synthesis. Users can spend more time using interactive texture synthesis for further refining the image. Thus, instead of looking at flat images, a user can fly through them after some simple processing. Minimal manual work, ease of use and interactivity are the salient features of our approach.     

Hybrid genetic optimization and statistical model based approachfor the classification of shadow shapes in sonar imagery

We present an original statistical classification method using a deformable template model to separate natural objects from man-made objects in an image provided by a high resolution sonar. A prior knowledge of the manufactured object shadow shape is captured by a prototype template, along with a set of admissible linear transformations, to take into account the shape variability. Then, the classification problem is defined as a two-step process: 1) the detection problem of a region of interest in the input image is stated as the minimization of a cost function; and 2) the value of this function at convergence allows one to determine whether the desired object is present or not in the sonar image. The energy minimization problem is tackled using relaxation techniques. In this context, we compare the results obtained with a deterministic relaxation technique and two stochastic relaxation methods: simulated annealing and a hybrid genetic algorithm. This latter method has been successfully tested on real and synthetic sonar images, yielding very promising results     

Scalable edge enhancement with automatic optimization for digital radiographic images

The ability to resolve fine picture detail is of paramount importance in medical imaging systems for viewing small tissue, bone structure and anatomy in X-ray images. In this paper, we present a new digital radiographic image processing system with the property of scalability and adaptability. (i) A new automatic optimization algorithm is proposed for display. (ii) An adaptive detection of a region-of-interest is developed. (iii) A “scalable edge enhancement algorithm” is proposed to improve the image quality for showing subtle structures in digital radiographic images. The advantage of the proposed method is demonstrated through experiments on 200 digital X-ray images and 50 CT images, in which different parts of human body structures are captured.     

Structure Aware Visual Cryptography

Visual cryptography is an encryption technique that hides a secret image by distributing it between some shared images made up of seemingly random black‐and‐white pixels. Extended visual cryptography (EVC) goes further in that the shared images instead represent meaningful binary pictures. The original approach to EVC suffered from low contrast, so later papers considered how to improve the visual quality of the results by enhancing contrast of the shared images. This work further improves the appearance of the shared images by preserving edge structures within them using a framework of dithering followed by a detail recovery operation. We are also careful to suppress noise in smooth areas.     

Dendritic stylization

Dendritic or branching structures are commonly seen in natural phenomena such as lightning, cracks, and vegetal growth. They are also often used for artistic or decorative purposes. We present a new procedural method for modeling dendritic structures based on a path planning approach. Our method includes the use of a partial non-scalar distance metric that gives us powerful and responsive control over the evolving dendritic structure. We demonstrate the effectiveness of our approach by creating dendritic stylizations of input images. We also show how our approach can be used to model more complex dendritic structures, such as trees; our algorithm allows us to create pareidolia effects, where an image is embedded within the branches of the tree.     

基于网格IC图象的多模板快速匹配算法

为了加快 IC图象中多个相似单元模板的匹配与定位 ,提出了一种基于网格 IC图象的多模板快速匹配算法 .该算法首先抽取网格图象和模板的二值拓扑结构 ,以构成图象和模板的粗分辨率表示 ;然后 ,在拓扑结构表示上通过综合来构造多模板的二叉树模型 ;接着 ,在二值拓扑结构表示上运用树模型进行搜索 ,在搜索过程中应用二叉决策树识别多个模板 ;最后 ,将粗匹配得到的目标 ,在原图象对应位置的小邻域内进行二次匹配 ,以确定模板和对应实例的位置 .应用此算法对 IC图象库进行测试 ,结果表明 ,所提出的多模板二叉决策树搜索算法与逐个模板匹配的方法相比 ,速度和效率均有较大幅度的提高     

风格强度可变的人脸风格迁移网络

目的 针对人脸风格迁移算法StarGAN （star generative adversarial network）、MSGAN （mode seeking generative adversarial network）等存在细节风格学习不佳、迁移效果单一和生成图像失真等缺点，提出一种能够降低失真并生成不同风格强度图像的人脸风格迁移算法MStarGAN （multilayer StarGAN）。方法 首先，通过特征金字塔网络（feature pyramid network，FPN）构建前置编码器，生成蕴含有图像细节特征的多层特征向量，增强生成图像在风格传输时能学习到的风格图像的细节风格；其次，使用前置编码器对原图像及风格图像各生成一个风格向量并进行组合，利用组合后的风格向量进行风格传输，使生成图像具有不同的风格迁移强度；最后，采用权重解调算法作为生成器中的风格传输模块，通过对卷积权重的操作代替在特征图上的归一化操作，消除特征图中的特征伪影，减少生成图像中的失真。结果 在Celeba_HQ数据集上进行实验，与MSGAN、StarGAN v2等对比算法相比，在参考引导合成实验中，MStarGAN的FID （Frechét inception distance score）指标分别降低了18.9和3.1，LPIPS （learnedperceptual image patch similarity）指标分别提升了0.094和0.018。在潜在引导合成实验中，MStarGAN的FID指标分别降低了20.2和0.8，LPIPS指标分别提升了0.155和0.92，并能够生成具有不同风格强度的结果图像。结论 提出的算法能够传输图像的细节风格，生成具有不同强度的输出图像，并减少生成图像的失真。     

A high‐level data fusion and spatial modelling system for change‐detection analysis using high‐resolution airborne digital sensor data

A high‐level data fusion system that uses Bayesian statistics involving weights‐of‐evidence modelling is described to combine disparate information from airborne digital data such as digital surface model (DSM), colour, thermal infrared (TIR) and hyperspectral images at different time periods. To determine the efficacy of the system, an analysis of change detection was performed. The data fusion system is capable of detecting changes in man‐made features automatically in a densely populated area where there is little prior information. Multiclass segmented images were obtained from the data captured by four airborne remote sensing sensors. The system performs data fusion modelling by using binary images of each theme class and a total of 40 binary patterns were obtained. Through Bayesian methods, involving weights‐of‐evidence modelling, all the binary images were analysed and finally four binary patterns (indicator images) were identified automatically as significant for the change‐detection application. A weighted index overlay model available in the system combines these four patterns. Data fusion by weights‐of‐evidence modelling is found to be straightforward and unequivocal for predicting newly transformed locations. The results of the Bayesian method are accurate as the weights are based on statistical analysis. Changes in features such as colour of roofs, parking areas, openland areas, newly built structures, and the presence or absence of vehicles are extracted automatically by using the high‐level data fusion approach. The final predictor image shows the probability of change‐detected areas in a densely populated city in Japan.     

基于空间结构统计建模的图像分类方法

提出一种基于图像空间结构统计建模的复杂纹理图像模式识别方法。从理论上分析了复杂纹理图像空间结构的韦伯分布过程,通过构造多尺度全向高斯导数滤波器,获得复杂纹理图像在不同观测尺度上的全方向空间结构统计建模表征结果。基于偏最小二乘-判决分析原理构建分类器,实现了复杂纹理图像的分类识别。实验结果表明,所提出的图像空间结构统计建模方法能获得复杂纹理图像关键性的视觉感知特性,基于该方法的图像分类准确率高且性能稳定。     

Image denoising using bilateral filter and Gaussian scale mixtures in shiftable complex directional pyramid domain

For image denoising, the main challenge is how to preserve the information-bearing structures such as edges and textures to get satisfactory visual quality when improving the signal-to-noise-ratio (SNR). Edge-preserving image denoising has become a very intensive research topic. In this paper, we describe a method for removing noise from digital images, based on bilateral filter and Gaussian scale mixtures (GSM) in shiftable complex directional pyramid (also named Pyramidal Dual-Tree Directional Filter Bank, PDTDFB) domain. Firstly, the noisy image is decomposed into different subbands of frequency and orientation responses using a PDTDFB transform. Secondly, the bilateral filter, which is a nonlinear filter that does spatial averaging without smoothing edges, is applied on the approximation subband. Finally, the distribution of detail subbands of PDTDFB coefficients is modeled with GSM, and the statistical model is then used to obtain the denoised detail coefficients from the noisy image decomposition by Bayes least squares estimator. Extensive experimental results demonstrate that our method can obtain better performances in terms of both subjective and objective evaluations than those state-of-the-art denoising techniques. Especially, the proposed method can preserve edges very well while removing noise.     

Reconstructing 3D human models with a Kinect

Three‐dimensional human model reconstruction has wide applications due to the rapid development of computer vision. The appearance of cheap depth camera, such as Kinect, opens up new horizons for home‐oriented 3D human reconstructions. However, the resolution of Kinect is relatively low, making it difficult to build accurate human models. In this paper, we improve the accuracy of human model reconstruction from two aspects. First, we improve the depth data quality by registering the depth images captured from multi‐views with a single Kinect. The part‐wise registration method and implicit‐surface‐based de‐noising method are proposed. Second, we utilize a statistical human model to iteratively augment and complete the human body information by fitting the statistical human model to the registered depth image. Experimental results and several applications demonstrate the applicability and quality of our system, which can be potentially used in virtual try‐on systems. Copyright © 2015 John Wiley & Sons, Ltd.     

Entropy Approach to Comparison of Images

Basically new pattern recognition method is implemented to compare two (or several) digital images. The method has neither feature alphabet, nor pattern dictionary recovery stages. It compares input images due to a special object called palette built from fragments of images. The measures to estimate the distances between images are based on a determination of the specific entropy of the frequency dictionary of an image with respect to the palette; that latter presents the statistical ancestor of the group of the images under comparison. The palette is defined as the frequency dictionary with frequencies of the fragments equal to arithmetic mean of the frequencies of the same fragments from the images to be compared. Such definition yields a minimum of the sum of specific entropies of the compared images with respect to the palette. Some preliminary results in the application of the method in pattern recognition and synergistics problems are presented. The limitations and basic properties of the method are discussed.     

数学形态法在超宽带SAR 道路边缘检测中的应用

超宽带SAR由于杂波模型多样化,当使用传统的基于单一杂波统计模型进行边缘检测时容易造成虚警高,定位精度差,边缘不连续,处理时间长等问题。基于数学形态学思想, 提出了一种多方向多尺度结构元素的二值形态学边缘检测算法,先对原始图像进行二值化处理,再运用多方向多尺度结构元素进行循环闭-开形态运算得到多个方向结果图,最后将各方向结果图进行融合得到最终的图像边缘。与传统边缘检测算法的对比实验表明,本文提出的算法由于采用了多尺度结构元素并且结合了图像的灰度信息,图像边缘检测虚警低,定位精确,耗时短,边缘更连续。     

流状线型结构图像修复算法的研究

针对流状线型结构图像修复问题,提出了一种基于张量扩散的流状结构图像修复算法,模型根据图像局部结构（局部图像结构由结构张量来度量）的纹理走向确定沿纹理方向和垂直纹理方向的扩散来修复断裂特征,并且控制沿纹理方向的扩散强度要大于沿垂直纹理方向的扩散强度,由此才能保证较好的纹理修复。为使计算结果更加准确,采用了非负性离散化和最优化旋转不变性两种数值计算方案。实验结果表明,对于有划痕或较小损坏区域的流状线型结构的纹理图像,该算法都能取得较好的修复结果。     

自适应尺度的双边纹理滤波方法

目前已有的结构保持的纹理平滑方法主要是利用矩形片内的统计量来区分纹理和结构,但是所用的矩形片边长是单一尺度的,这将导致含有尖锐结构或结构在多个尺度上的图像出现纹理过平滑或未平滑的现象。为此,提出一种自适应尺度的双边纹理滤波方法。首先,通过对局部区域进行统计分析,从给定候选值中自适应地为每个像素选取合适的矩形片边长,对于均匀的纹理区域,选取较大的矩形片边长,对于邻近特征边的区域选取较小边长;其次,利用自适应的矩形片边长计算引导图像;最后,对原始图像进行引导双边滤波。实验结果表明,所提方法能够在保持图像结构的同时更好地平滑纹理。     

基于曲波变换的红外/可见光图像融合

和小波变换相比较,曲波变换能更好地表示图像的边缘信息.在此基础上给出了一种基于曲波变换的图像融合方法,并将其应用于红外和可见光图像融合.首先,对红外图像和可见光图像分别进行曲波变换,得到两幅图像的低频分量和不同尺度的高频分量.在对源图像的各分量融合时,对低频分量采用平均加权进行融合.对高频分量采用取绝对值较大的方法进行融合,得到融合后的低频分量和不同尺度的高频分量,最后对这些融合后的分量进行重构,得到融合图像.仿真结果表明:和基于小波变换的融合算法相比较,该算法较好地保留了源图像的细节信息,提高了融合的效果.     

基于小波域边缘方向特征的SAR图象噪声抑制方法

给出了一种新的基于小波变换的合成孔径雷达 (SAR)图象斑点噪声抑制方法 .利用每一级小波分解得到的小波系数子带 HL和 L H,以及对原图进行水平方向旋转正负 4 5°扫描后得到的另外两个正交方向的小波系数子带 rc HL和 ra HL ,可以判断出对应点边缘方向性的强弱 ,通过设定方向性阈值 ,确定该点是否位于边缘上 ,进而对没有位于边缘的点进行平滑 ,达到保留图象边缘的同时 ,抑制斑点噪声的目的 .为解决对某些振荡型边缘的检测问题 ,还结合阈值法 ,对该方法做了改进 .实验表明 ,与小波域的硬阈值或软阈值去噪方法相比 ,此方法在有效地抑制斑点噪声的同时 ,更好地保留了 SAR图象中的边缘和纹理信息 .     

Scale‐aware Structure‐Preserving Texture Filtering

This paper presents a novel method to enhance the performance of structure‐preserving image and texture filtering. With conventional edge‐aware filters, it is often challenging to handle images of high complexity where features of multiple scales coexist. In particular, it is not always easy to find the right balance between removing unimportant details and protecting important features when they come in multiple sizes, shapes, and contrasts. Unlike previous approaches, we address this issue from the perspective of adaptive kernel scales. Relying on patch‐based statistics, our method identifies texture from structure and also finds an optimal per‐pixel smoothing scale. We show that the proposed mechanism helps achieve enhanced image/texture filtering performance in terms of protecting the prominent geometric structures in the image, such as edges and corners, and keeping them sharp even after significant smoothing of the original signal.