融合Hough直线检测和Grab-cut的风机叶片自适应分割方法

完整的边缘信息对风力发电机叶片的边缘缺陷检测至关重要,但由于户外采集的风机叶片图像背景复杂多样,现有图像分割算法的分割精度不足,无法保证边缘缺陷的完整性.因此,提出一种基于图像边缘特征与颜色信息的自适应图像分割方法实现风机叶片边缘检测.首先,使用Hough直线检测初步定位叶片直线边缘;然后,在目标区域应用基于Otus阈...     

基于改进Census变换的自适应局部立体匹配

针对现阶段局部立体匹配在弱纹理区域具有匹配精度低且过度依赖中心像素的缺点,提出一种基于改进Census变换的自适应局部立体匹配算法。首先根据中心像素领域的纹理复杂度采用自适应支持窗口改进Census变换,引入Tanimoto系数与Hamming距离算法结合,并融合颜色或亮度差的绝对值用作新的初始匹配代价计算。通过十字交叉域算法进行代价聚合并采用赢家通吃算法计算视差,在视差优化阶段采用左右一致法、迭代投票、插值填充和亚像素细化,针对边缘模糊化将改进的自适应中值滤波用作抑制噪声得到最后的视差图。实验结果表明,本文所提出的算法在Middlebury数据集上的平均误匹配率为4.39%,相较于其他改进的Census变换算法有明显提升,并在抗噪能力上具有一定的鲁棒性和适应性。     

小波包结点阈值自适应消噪法

当噪声为时变的时候,传统的小波阈值消噪方法效果很有限:当采用软阈值消噪时,总体效果较好,但当含噪信号很不规则时显得过于光滑;当采用硬阈值消噪时,消噪效果不理想,信号含有明显的噪声。为克服上述缺陷,本文提出基于"自适应阈值消噪思想"的小波包结点阈值消噪法。首先,我们给出了小波包结点阈值的定义,这种阈值取法突破了高斯白噪声和平稳噪声的局限性;其次,我们提出了基于谱熵的噪声估计法用以替代传统的中值绝对值(MAD)噪声估计法,这种噪声估计方法与实际应用环境相符,适用于非平稳噪声和有色噪声。仿真实验证实,相比于常规小波消噪算法,结点阈值法和基于谱熵的噪声估计法在白噪声环境下具有较好的消噪效果。     

车道线检测的PSPNet改进算法

车道线检测已成为智能驾驶领域研究的一项重要课题,而实际应用时,常出现车道线分割不准确、实时检测能力不佳的问题。为此本文提出一种金字塔场景分析网络的改进算法。在编码结构的基础上搭建主体网络PSPNet,选用MobileNet v2轻量级网络作为编码器的主干网络,减少了整体网络的计算复杂度及参数量;网络中添加了空洞卷积,并在不同层间实现特征融合,扩充了模型感受野,同时丰富了局部特征;最后用自适应直线拟合算法对各类型车道线拟合。本文使用Caltech车道线数据集进行测试,实验结果显示,改进后的PSPNet算法对不同类型的车道线均有较好的分割结果,与PSPNet算法相比精度和交并比分别提升了3.91%、4.14%,且FPS达28帧/s,本文算法的分割精度和推理速度均优于其他对比算法。     

基于POBT的空瓶规则纹路区域缺陷检测

现有的玻璃瓶瓶身质量检测方法中,缺乏对有纹理图案的玻璃瓶的缺陷检测,针对这一问题,提出一了种仅基于相位变换(phase only based transition,POBT)的玻璃瓶纹理区域缺陷检测算法。该算法采用基于灰度投影的方法求取玻璃瓶瓶身中心坐标,利用基于相位的变换方法,通过归一化过程去除玻璃瓶规则纹路区域的低频纹路分量,仅保留下缺陷图像,应用概率修正自适应阈值分割方法对经过POBT变换后去除纹理留下的缺陷的图像进行分割,提取缺陷,并将该分割方法与3种传统型的方法进行对比。实验结果表明,该方法可实现瓶身规则纹路区域的高速高精度缺陷检测。     

Color-based road detection in urban traffic scenes

Road detection is a key issue for autonomous driving in urban traffic. In this paper, after a brief overview of existing methods, we present a road-area detection algorithm based on color images. This algorithm is composed of two modules: boundaries are first estimated based on the intensity image and road areas are subsequently detected based on the full color image. In the first module, an edge image of the scene is analyzed to obtain the candidates for the left and right road borders and to delimit the area that will subsequently be used to compute the mean and variance of the Gaussian distribution, assumed to be obeyed by the color components of road surfaces. The second module effectively extracts the road area and reinforces boundaries that most appropriately fit the road-extraction result. The combination of these modules can overcome basic problems due to inaccuracies in edge detection based on the intensity image alone and due to the computational complexity of segmentation algorithms based on color images. Experimental results on real road scenes have substantiated the effectiveness of the proposed method.     

Underwater Image Enhancement Based on IMSRCR and CLAHE-WGIF

Aiming at the scattering and absorption of light in the water body, which causes the problems of color shift, uneven brightness, poor sharpness and missing details in the acquired underwater images, an underwater image enhancement algorithm based on IMSRCR and CLAHE-WGIF is proposed. Firstly, the IMSRCR algorithm proposed in this paper is used to process the original underwater image with adaptive color shift correction; secondly, the image is converted to HSV color space, and the segmentation exponential algorithm is used to process the S component to enhance the image saturation; finally, multi-scale Retinex is used to decompose the V component image into detail layer and base layer, and adaptive two-dimensional gamma correction is made to the base layer to adjust the brightness unevenness, while the detail layer is processed by CLAHE-WGIF algo-rithm to enhance the image contrast and detail information. The experimental results show that our algorithm has some advantages over existing algorithms in both subjective and objective evaluations, and the information entropy of the image is improved by 6.3% on average, and the UIQM and UCIQE indexes are improved by 12.9% and 20.3% on average.     

Fast Image Segmentation Algorithm Based on Salient Features Model and Spatial-frequency Domain Adaptive Kernel

A fast image segmentation algorithm based on salient features model and spatial-frequency domain adaptive kernel is proposed to solve the accurate discriminate objects problem of online visual detection in such scenes of variable sample morphological characteristics, low contrast and complex background texture. Firstly, by analyzing the spectral com-ponent distribution and spatial contour feature of the image, a salient feature model is established in spatial-frequency domain. Then, the salient object detection method based on Gaussian band-pass filter and the design criterion of adaptive convolution kernel are proposed to extract the salient contour feature of the target in spatial and frequency domain. Finally, the selection and growth rules of seed points are improved by integrating the gray level and contour features of the target, and the target is segmented by seeded region growing. Experiments have been performed on Berkeley Segmentation Data Set, as well as sample images of online detection, to verify the effectiveness of the algorithm. The experimental results show that the Jaccard Similarity Coefficient of the segmentation is more than 90%, which indicates that the proposed algorithm can availably extract the target feature information, suppress the background texture and resist noise interference. Besides, the Hausdorff Distance of the segmentation is less than 10, which infers that the proposed algorithm obtains a high evaluation on the target contour preservation. The experimental results also show that the proposed algorithm significantly improves the operation efficiency while obtaining comparable segmentation performance over other algorithms.     

复杂背景下多特征结合的深度学习手势识别

科学技术的快速发展使得基于深度学习的人机交互已经得到广泛的应用。手势识别作为人机交互领域的重要组成部分,同样具有重要的研究意义和应用价值。通过对传统的手势识别方法进行研究,发现主要是利用肤色检测算法实现手势识别和分类,但是传统方法在针对具有复杂背景的手势图像时会出现识别效果差等问题。为解决这一问题,提出一种基于卷积神经网络的肤色特征和边缘特征结合的手势识别方法。首先,在YCrCb颜色空间采用椭圆肤色模型和Otsu阈值肤色识别算法获取手势肤色特征,经算法判断后,对手势肤色图像采用改进Canny边缘检测算法获得手势边缘特征。其次,提出一种边缘填充方法对手势边缘图像处理,得到手势轮廓完整的手势边缘图像。最终,采用逻辑运算和形态学运算得到手势分割图像,并输入卷积神经网络进行训练和识别。实验结果表明,该方法在复杂背景下具有较好的手势识别效果,在NUS-II数据集上的平均识别率为98.83%。     

Automatic color segmentation algorithms-with application to skintumor feature identification

Two color-image segmentation methods are described. The first is based on a spherical coordinate transform of original RGB data. The second is based on a mathematically optimal transform, the principal components transform (also known as eigenvector, discrete Karhunen-Loeve, or Hotelling transform). These algorithms are applied to the extraction from skin tumor images of various features such as tumor border, crust, hair scale, shiny areas, and ulcer. The results of this research will be used in the development of a computer vision system that will seve as the visual front-end of a medical expert system to automate visual feature identification for skin tumor evaluation     

改进的Yoon自适应加权立体匹配算法

分析了Yoon自适应加权立体匹配算法在权值计算与像素相异性度量上所存在的问题,并对其进行了改进。首先分析了Yoon自适应加权立体匹配算法颜色相似性和空间邻近性对权值(视差)的不同影响,据此提出新的空间临近性权值影响模型;然后将Birchfield相异性度量函数从2-邻域扩展到8-邻域,消除行采样对像素相异性度量的影响,使得匹配代价计算更为准确;最后通过对称加权确定视差。实验表明,新方法改善了Yoon算法在深度不连续区,低纹理区,高纹理区以及重复结构区的匹配性能,可获得更为精密的视差图。     

Morphometric analysis of trabecular bone thickness using different algorithms

Investigations have been carried out with the goal of assessing the trabecular bone thickness of biological samples using images obtained by micro-computed tomography and magnetic resonance imaging. There is no conventional definition of trabecular bone thickness, and many methods may be involved in determining it. However, the results of the available algorithms or software packages differ considerably from each other. This paper determines trabecular bone thickness on the basis of several algorithms. A deep understanding of the performance of different methods is achieved by studying pseudo-three-dimensional images of both geometrical models of well-defined thickness and real bone samples with different bone densities. The models facilitate comparisons between the algorithms or software packages. Comparison of the results obtained from these commercial software packages and other state-of-the-art algorithms shows that the thickness, spatial distribution, and shape of an object affect each result differently, but in a significant manner. This is primarily due to variations in the thresholding algorithms used to distinguish object area elements (pixels/voxels) from the background, or non-object, region. Additionally, the results show that the average difference in thickness measurements can vary by up to 102.34% for models and 46.49% for real bone samples. This data shows that the differences in measurements of the trabecular bone thickness due simply to the algorithm involved are remarkable. Therefore, biomedical engineers and scientists should be careful to select the algorithm that is most compatible with their specific application.     

基于改进 KCF 算法和多特征融合的车辆跟踪研究

针对目前车辆跟踪研究算法中,核相关滤波算法(KCF)在复杂背景下存在特征提取单一以及尺度无法自适应的不足,本文提出一种多特征融合的尺度自适应算法。该算法以颜色直方图信息作为颜色特征,将具有更多语义信息的高层卷积特征和拥有较高分辨率的底层卷积特征作为深度特征,并与颜色特征进行自适应特征融合。然后,采用上下文图像对目标背景信息进行约束优化,并通过平均峰值相关能量检测衡量响应置信度,最后利用高置信度的跟踪结果来避免模型易受干扰的问题。通过在OTB100数据集上的实验表明,本文算法的精度分别比其他的主流跟踪算法Staple、SAMF、KCF、TLD、DSST和CSK高出4.9%, 5.7%, 10.2%, 10.3%, 23.4%, 29.7%。     

基于数字图像处理的液位测量系统的研究与实现

基础设备的数字化是智慧电厂发展的基本要求。为实现基地式液位仪表的远程监视与自动读数,提出了一种基于数字图像处理的液位测量系统。图像处理算法包括颜色阈值分割、改进的Canny边缘检测、模板匹配,并提出窗口搜索峰值检测算法,分别对采集图像进行仪表定位、液位分界面提取、数字识别和摄像头自标定,从而实现液位图像的数字化。同时,该测量系统具有自适应中值滤波、直方图分析和透视失真自矫正功能,增强了测量的抗干扰性和自适应性。经实验室试验,测量系统的有很好的准确性,能维持原有仪表的精度等级。     

基于多通道增强融合的自适应液晶屏图像分割

液晶屏电子仪表因其示数精确,抗干扰能力强的特性,逐渐取代了指针式仪表,在工业生产中得到了广泛的应用。 针对 液晶屏的自身的显示特性和显示原理,提出一种利用信息熵进行自适应彩色形态学预处理且基于 HSV 空间和色调映射 V 通道 增强液晶屏区域分割方法,并结合了双通道图像加权融合的区域图像增强算法研究,实现液晶屏位置区域的分割。 对多种液晶 屏图像的分割实验的结果表明,较之传统液晶屏分割方案,该方法对单色液晶屏的分割准确度提升 8. 6%,对彩色液晶屏分割准 确度提升了 12. 4%。     

彩色复制中的色域映射

由于各种颜色设备能表现的颜色范围，即色域各不相同，用一个设备复制另一个设备的颜色时，可能无法原样再现。为了尽可能减少这种颜色失真，在复制前需进行设备间的色域映射操作，本文举例介绍了当前颜色复制系统中的色域映射技术和算，分析比较了各种方法及色域映射中的相关问题。     

基于改进LCCP的堆叠物体分割算法

局部凸连接生长算法（LCCP）存在超体素跨越物体边界,未能利用区域隐含凹凸信息的缺陷,为了改进以上缺陷导致的分割精确度低、物体粘连的问题,提出了结合连通域分割的改进算法。首先采用深度自适应超像素分割法（DASP）根据深度信息和法向量角度将图片划分为超像素;其次根据超像素的法向量夹角判定邻接超像素的凹凸性,合并所有凸连接超像素形成初步结果;最后使用基于超像素的距离变换以及分水岭生长分割方法,把面积较大的凹连通域,快速分割成多个凸区域。在IC-BIN数据集进行分割验证,结果表明平均分割精度（AP）相比于LCCP和约束平面切割法（CPC）分别提升25%和35%,显著改善了欠分割问题。     

自适应粒子群集优化二维OSTU的图像阈值分割算法

为了解决红外相机采集行人图片时图像分割效果问题,提出一种自适应粒子群优化二维OSTU的阈值分割算法。利用当前帧图像的灰度级和当前帧图像像素的邻域灰度级构成二元组,通过计算二者的均值和方差,建立二维最大类间方差模型,结合自适应粒子群集算法,估计出图像的最佳阈值,该方法不仅能够准确估计阈值且计算时间减少。仿真结果表明,阈值最佳时,当结合自适应粒子群集优化算法后计算时间减少到原来的50%,所提出的算法能够快速准确得到最佳阈值,提高了图像预处理的分割效果。     

Maximum likelihood-based adaptive differential evolution identification algorithm for multivariable systems in the state-space form

Parameter estimation plays an important role in the field of system control. This article is concerned with the parameter estimation methods for multivariable systems in the state-space form. For the sake of solving the identification complexity caused by a large number of parameters in multivariable systems, we decompose the original multivariable system into some subsystems containing fewer parameters and study identification algorithms to estimate the parameters of each subsystem. By taking the maximum likelihood criterion function as the fitness function of the differential evolution algorithm, we present a maximum likelihood-based differential evolution (ML-DE) algorithm for parameter estimation. To improve the parameter estimation accuracy, we introduce the adaptive mutation factor and the adaptive crossover factor into the ML-DE algorithm and propose a maximum likelihood-based adaptive differential evolution algorithm. The simulation study indicates the efficiency of the proposed algorithms.