The fog density level, as one of the indicators of weather conditions, will affect the management decisions of transportation management agencies. This paper proposes an image-based method to estimate fog density levels to improve the accuracy and efficiency of analyzing fine meteorological conditions and validating fog density predictions. The method involves two types of image entropy: a two-dimensional directional entropy derived from four-direction Sobel operators, and a combined entropy that integrates the image directional entropy and grayscale entropy. For evaluating the performance of the proposed method, an image test set and an image training set are constructed; and each image is labeled as heavy fog, moderate fog, light fog, or fog-free according to the fog density level of the image based on a user study. Using our method, the average accuracy rates of image fog level estimation were 77.27% and 79.39% on the training set using the five-fold cross-validation and the test set, respectively. Our experimental results demonstrate the effectiveness of the proposed combined entropy for image-based fog density level estimation.

To reduce the flicker artifacts caused by video defogging, a surveillance video defogging algorithm based on the background extraction and consistent constraints is proposed. First, an inter frame consistency constraint is constructed and applied to background modeling. Second, the extracted background is defogged with an improved static defogging approach. Third, the foreground is extracted using the extracted background and further defogged using constraints of the consistency between the foreground and background. Experimental results show that our algorithm can remove fog effectively and preserve the temporal coherence well.

This paper proposes a method and an original index for the estimation of fog density using images or videos. The proposed method had the advantages of convenient operation and low costs for applications in automatic driving and environmental monitoring. The index was constructed based on a dark channel map and the pseudo-edge details of the foggy image. The effectiveness of the fog density index was demonstrated and validated through experiments on the two existing open datasets. The experimental results showed that the presented index could correctly estimate the fog density of images: (1) the estimated fog density value was consistent with the corresponding label in the Color Hazy Image Database (CHIC) in terms of rank order; (2) the estimated fog density level was consistent with the corresponding label in the Cityscapes database and the accuracy reached as high as 0.9812; (3) the proposed index could be used to evaluate the performance of a video defogging algorithm in terms of residual fog.

为了提高图像去雾效果,提出一种在图像多尺度分解的基础上,通过改进的暗原色方法实现图像去雾的算法.该算法在HSV颜色空间中,首先对亮度分量V进行多尺度分解,然后对高频部分进行软阈值消噪,并通过梯度运算增强边缘,对低频部分采用改进的暗原色先验模型进行处理;其次,对饱和度分量S先进行非局部均值滤波去噪,再通过线性拉伸使其复原;最后,将图像由HSV空间转换到RGB空间得到最终的去雾图像.实验结果表明,在去雾效果和时间复杂度方面,该算法比一些经典算法都有明显的改善.

提出了可协调经验小波变换,并将其应用于多聚焦图像融合.经验小波变换(EWT)是一种自适应信号分解方法,具有比经验模态分解和传统小波分解更好的特性.其核心思想是通过构造自适应的滤波器实现对信号的自适应分解.但是若直接对两幅多聚焦图像分别进行EWT分解,因各自生成的经验小波互不相关,将出现分解所得对应子带不匹配的情况,影响融合图像的质量.针对这一问题,提出了一种可协调的经验小波变换(C-EWT),C-EWT分解下的两个多聚焦图像的对应子带是完全匹配的.基于此,利用C-EWT提出了一种新的多聚焦图像融合算法.每幅源图像经过C-EWT分解后,得到一个低频分量和多个高频分量;对低频分量采用基于改进Laplacian能量和的阈值匹配选择与加权规则进行融合,对高频分量则采用局部Log-Gabor能量取大的融合规则;将融合之后的各子带分量进行重构得到融合图像.仿真实验表明:与其他六种融合算法相比,所提算法在融合聚焦区域、保留边缘和细节信息方面具有优势,融合图像具有更好的视觉效果,且客观评价指标与标准图像最为接近.

Based on image segmentation and the dark channel prior, this paper proposes a fog removal algorithm in the HSI color space. Usually, the dark channel prior based defogging methods easily produce color distortion and halo effect when applied on images with a large sky area, because the sky region does not meet the prior assumption. For this reason, our method presents a new threshold sky region segmentation algorithm using the initial transmission map of the intensity component I.Based on the segmentation result, the initial transmission map is modified in turn, and finally refined by the guided filter. The saturation components S is reconstructed using the low frequencies of the V-transform to reduce noise, and stretched by multiplying a constant related to the initial transmission map. Experimental results show that the proposed algorithm has low time complexity and compelling fog removal result in both visual effect and quantitative measurement.

Single image dehazing algorithm based on the dark channel prior may cause block effect and color distortion. To improve these limitations, this paper proposes a single image dehazing algorithm based on the V-transform and the dark channel prior, in which a hazy RGB image is converted into the HSI color space, and each component H, I and S is processed separately. The hue component H remains unchanged, the saturation component S is stretched after being denoised by a median filter. In the procession of intensity component, a quad-tree algorithm is presented to estimate the atmospheric light, the dark channel prior and the V-transform are used to estimate the transmission map. To reduce the computational complexity, the intensity component I is decomposed by the V-transformfirst, coarse transmission map is then estimated by applying the dark channel prior on the low frequency reconstruction image, and the guided filter is finally employed to refine the coarse transmission map.For images with sky regions, the haze removal effectiveness can be greatly improved by just increasing the minimum value of the transmission map. The proposed algorithm has low time complexity and performs well on a wide variety of images. The recovered images have more nature color and less color distortion compared with some state-of-the-art methods.

目的 针对传统的基于多尺度变换的图像融合算法的不足,提出了一种基于W变换和2维经验模态分解(BEMD)的红外与可见光图像融合算法.方法 首先,为了更有效地提取图像的高频信息,抑制BEMD中存在的模态混叠现象,提出了一种基于W变换和BEMD的新的多尺度分解算法(简称W-BEMD);然后,利用W-BEMD对源图像进行塔式分解,获得图像的高频分量WIMFs和残差分量WR;接着,对源图像对应的WIMFs分量和WR分量分别采用基于局部区域方差选择与加权和基于局部区域能量选择与加权的融合规则进行融合,得到融合图像的W-BEMD分解;最后,通过W-BEMD逆变换得到最终融合图像.W-BEMD分解算法的主要思想是通过W变换递归地将BEMD分解过程中每层所得低频分量中滞留的高频成分提取出来并叠加到相应的高频分量中,实现更有效的图像多尺度分解.结果 对比实验结果表明,本文方法得到的融合图像视觉效果更佳,既有突出的红外目标,又有清晰的可见光背景细节,而且在平均梯度(AG)、空间频率(SF)、互信息(MI)3个客观评价指标上也有显著优势.结论 本文提出了一种新的红外与可见光图像融合算法,实验结果表明,该算法具有较好的融合效果,在保留可见光图像中的细节信息和突出红外图像中的目标信息方面更加有效.

针对光照不均匀的图像,结合W系统和NSCT变换,提出了一种新的图像增强方法.方法首先利用W变换对图像进行多尺度分解,然后利用NSCT中的非下采样方向滤波器组对尺度分解中的高频部分进行方向分解,得到不同尺度不同方向上的变换系数.在多尺度几何分解的基础上,对低频子带图像采用动态直方图均衡化、高频子带图像采用同态滤波的方法进行增强处理,最后利用非线性函数减小图像明、暗部分灰度值的差异,得到最后的增强结果.仿真实验结果表明,算法无论在视觉效果上还是客观评价指标上都优于其他被比较的四种增强算法,对于过亮、过暗以及局部光照不均匀的图像均取得了更好的增强效果,在增强图像细节的同时能有效抑制图像的伪吉布斯失真和过增强失真.在评价指标上,算法对三组经典图像处理后的增强图像的信息熵分别达到了10.0755、9.7879、10.5338,明显优于其他方法.

特征值与特征向量是线性代数中的2个重要概念,在科学研究和工程技术中有着广泛的应用.特征值与特征向量的概念抽象难懂,直观引入和应用实例的融入能让学生更好地理解概念的本质.运用数形结合的方法,从线性不变量入手引出特征值与特征向量,并结合应用实例激发学生的学习热情,让学生在分析和解决问题的过程中加深对概念的理解,同时提高应用知识解决实际问题的能力.

为提高包含天空区域图像的去雾效果, 解决暗原色先验原理去雾容易导致天空区域色彩失真以及图像整体亮度较暗的问题, 针对含有大片天空的图像, 提出一种基于天空分割和色调映射的图像去雾算法. 在 HSI 颜色空间中利用图像众数和图像连通区域提出天空识别算法, 分割出天空与非天空区域; 然后根据暗原色先验原理分别求取二者透射率, 并在天空区域完成大气光值的估计; 最后在 RGB 空间中利用大气散射模型复原图像, 并经过改进的自适应色调映射得到最终的去雾图像. 采用合成雾图、实景雾图和网络收集雾图进行实验的结果表明, 该算法在主观视觉和客观指标方面均能得到质量更好的去雾图像.

To explore the effective mathematical tools (orthogonal function and orthogonal transformation) for signal analysis more deeply, a new orthogonal functions system composed of piecewise polynomials is constructed from the Legendre polynomials, and it is called generalized V-system of degree k. Equivalence relation between the generalized V-system of degree k and the V-system of degree k is proved. The discrete matrix corresponding to the generalized V-system of degree k is built, by which the generalized V-transform of degree k is constructed. It is proved that the generalized V-transform maintains almost all the properties of the V-transform, and also has a fast algorithm for degree one, which makes it superior to the V-transform as designing fast algorithms for the V-transform is difficult. The experimental results show that the time effi-ciency of the fast generalized V-transform is substantially increased compared with the V-transform.

To eliminate the influence of foggy weather on image acquisition,improve image quality and solve the problem of incomplete retention of image information and poor articulation for traditional dehazing techniques,this paper proposes an improved dark channel prior based image dehazing algorithm using color space conversion.The RGB color space of the image is first converted to the HSI color space,while the hue component H remains unchanged.Then we perform dark channel prior based image dehazing on the intensity component Ⅰ,and simultaneously calculate the atmospheric light value using the more accurate quad-tree algorithm.The saturation component S is performed by V-transform,and the low-frequency saturation is reconstructed to reduce the influence of texture and noise,and to increase saturation.For images containing large areas of the sky,fog and haze can be effectively removed by further improving the minimum transmittance,and color distortion can be also avoided in this process.Experiment results show that the algorithm has obvious haze removal effect,the dehazed image has higher clarity and more realistic color,retains relatively complete image information,and has lower time complexity compared with the classic haze removal algorithm.

A multi-wavelet system with detailed mathematical expression called V-system is introduced.The infrared and visible images are decomposed into different layers and orientations by using multi-resolution of V-system and multi-orientation in non-subsampled contourlettransform(NSCT).And then different fusion strategies were adopted to fuse raw images in each layer and each orientation respectively.Firstly,the original image was decomposed by multilevel V-decomposition,and contour informations and multi-layer detailed informations of images were gotten;then the obtained contour informations were decompose again by NSCT to obtain low frequency and high frequency coefficients.The low frequency coefficients are fused according to the strategy based on sparse representation,and the high frequency coefficients are fused according to the strategy based on 2D Log-Gabor energy,and then the improved pulse coupled neural network was used to fuse multi-layer detailed information.Finally,the fused image is obtained by the corresponding inverse transformation.The algorithm decomposes images in different layers and orientations to obtain more refined detail of raw images.The combination of various fusion strategies makes the detailed information more clear and enhances the contrast of the fused images,and it also improves the objective indicators observably.

Multi-scale decomposition method and fusion rule are two key factors for multi-focus image fusion method.The W-system of degree k is an orthogonal hybrid function system consisting of polynomials of degree k and piecewise polynomials of degree k.The corresponding W-transform is an effective multi-resolution analysis tool with orthogonality and reproducibility.By combining the multiscale characteristics of W-transform and multidirection property of the non-subsampled directional filter bank (NSDFB) transform, a new multi-scale and multi-directional transform is proposed in this paper, in which multi-scale decomposition is performed by the W-transform and the multi-directional decomposition is obtained from NSDFB.On this basis, a new multi-focus image fusion algorithm is proposed.According to the characteristics of the high frequency coefficients, the often used simplified PCNN is modified, and applied on the fusion rules of the high frequency coefficients.The experimental results show that the proposed fusion method can select pixels from focus areas of the source images and effectively avoid generation of artifacts.The fused image has better visual effect.The objective evaluation index including standard deviation, entropy, average gradient, and spatial frequency of the fused image demonstrate that the proposed method outperforms the fusion methods based on Contourlet transform, NSCT transform and discrete wavelet.

Infrared and visible image fusion can produce a composite image which has high contrast and rich background details of the scene. In view of the defects of some existing infrared and visible fusion method, such as low contrast and unclear background details, we propose a novel multi-scale fusion method based on the combination of non-sampled contourlet transform (NSCT), sparse representation and pulse coupled neural network. In our method, the source images are firstly decomposed into one low frequency sub-band and high frequency sub-bands at different scales and directions using NSCT. Fusion rules based on the sparse representation and modified PCNN are developed, and then used for fusion of the low sub-band and high frequency sub-bands, respectively. In the modified PCNN developed in this paper, we use Sum-Modified-Laplacian and Log-Gabor energy as values of the linking strength instead of setting it a constant. Each of the linking strength corresponds to an ignition map, the average of the two results is taken as the final PCNN output. The fused image are finally obtained by performing the inverse NSCT. Comparison experiment results show that the fused image produced by the proposed method has high contrast and rich details, as well as the greatly improved objective evaluation indexes values.

Fused images obtained using the traditional multi-modal medical image fusion technology cannot express details clearly and lesion obviously.In view of this,a new fusion method which combines the V-transform and Nonsubsampled Contourlet Transform(NSCT) is proposed.The source images are first decomposed into contour sub-image and detail sub-images by applying the multi-layer V-decomposition,and then NSCT transform is performed on the contour sub-image.Fusion rule in NSCT domain is designed.Fusion strategy for detail information is presented on detail sub-images.The fused image is finally obtained by overlaying the fused contour image and fused detail image together.Experimental results show that the proposed algorithm outperforms the traditional discrete wavelet transform and NSCT transform in both visual effect and evaluation indexes.

To improve the quality of low illumination image, a new image enhancement algorithm based on HSI color space is proposed. First, we convert the RGB image into the HSI color space, and then perform enhancement to the saturation S and brightness I respectively with different methods. To make the image more suitable for human visual habit, piecewise exponential transformation is used for the enhancement of the saturation component S. While the V-system, a new orthogonal multi-wavelet, is introduced for the enhancement of the luminance component I. First, we perform V-transform to the luminance I to isolate the high, low frequency sub-bands. And then process Retinex adjustment to the low-frequency sub-band to reduce the influences of illumination to the image, while we use improved fuzzy enhancement to the high frequency sub-band to achieve image denoising and enhancement. Finally, the processed S, I and H components are synthesized into a clear RGB image. The experiment results show that our algorithm can obviously improve the visual effect of the low illumination color image, and also has significant improvement in the objective evaluation indexes.

随着部分高等农林院校向综合性大学发展,数学硕士学位研究生的培养在高等农林院校逐步得以开展.目前,高等农林院校数学硕士学科点的建设主要具有创建时间短、基础较弱但是教师集体凝聚力强、研究方向的互补性强、具有较多的交叉研究机会等特点;同时,面临着师资力量有待加强,研究方向分散、缺乏特色,生源数量不足、素质参差不齐,本、硕专业的关系有待理顺,学科点内外的有效学术交流有待加强等问题.为此,提出高等农林院校数学硕士学科点建设应侧重于2个关键点:一是通过引进优秀人才和优化师资结构,完善导师遴选机制、考核机制和激励机制,加强导师的职前和在岗培训,保障导师的科研工作时间等,加强导师队伍建设;二是通过加强基础数学领域相关研究方向的建设、加强与农林类相关学科的合作研究、依托所在区域的数学学科优势、在短时间内凝练出主干研究方向等,加强研究方向建设.同时,从课程建设、人才培养模式、交流合作、理顺本、硕专业的关系等方面,提出进一步加强高等农林院校数学硕士学科点建设的建议.

Image understanding is based on image semantic segmentation. By employing patch shape complexity index and Euler number, an image segmentation algorithm ( GM ) based on over-segmentation, patch merging and graph model is presented. Following a common technical tactic that an image is segmented after smoothing, 16 algorithms for image segmentation are presented by combining each of the 4 filter methods including Bilateral Filtering ( BiF ) , Edge Preserving Smoothing (EPS), structure extraction via Relative Total Variation (RTV) and Rolling Guidance Filter ( RGF ) , with each of the 4 segmentation algorithms including Threshold Segmentation ( ThrS) , Active Contours without Edges ( ACnE) , Particle Swarm Optimization ( PSO) , and GM. To evaluate the efficiency of the algorithms, the comparative experiments are conducted, and two evaluation indexes, Corrected Rand Index ( CRI ) and Jaccard Index ( JI ) , are used in the experiments. The experimental results show that the algorithm which segments an image with GM after RTV has the best performance from CRI and JI. T-test also shows that it is statistically significant that the segmentation effect of GM algorithm is improved by using image smoothing.