用于大豆品种识别的叶片深度特征学习方法

大豆有许多品种(cultivar),它们的叶片图像模式的差异非常细微,因此很难通过叶片特征将大豆品种区分开.虽然在使用叶片图像模式进行植物种类(species)识别方面的研究已经取得了巨大的进步,然而,作为一项非常细粒度的模式识别问题,大豆品种的识别与分类研究尚未引起足够的重视.传统的手工叶片图像分析方法一般无法刻画不同大豆品种的叶片特征的细微差异,因此识别率很低.本文尝试使用深度学习来提取具有强的辨识能力的叶片特征,以解决大豆的品种识别问题.我们提出了一种新颖的深度学习模型,称为目标转换注意力网络(Transformation Attention Network,TAN).该方法首先通过注意力机制提取细粒度的叶片图像特征,然后使用仿射变换纠正叶片姿势.我们构建了一个由240个大豆品种组成的大豆叶片品种图像数据库,每个品种有10个样本,以此数据集验证叶片图像模式中品种信息的可用性,并验证了所提出的深度学习模型对大豆品种识别的有效性.令人鼓舞的是实验结果证实了叶片图像模式在区分栽培大豆品种方面的有效性,并证明了所提出的方法优于流行的叶片手工特征提取方法和深度学习方法.     

基于叶片形态特征的葡萄品种自动识别

葡萄在全世界种植范围较广,葡萄品种识别对葡萄资源统计、新品种鉴定及遗传资源保护都有重要意义。以成熟叶片为基本材料,不变矩、面积、周长、主叶脉长度和叶柄长度等11维特征向量为基础数据,采用叶片特征选择、图像处理、特征值提取、建立分类模型等方法,运用国际葡萄与葡萄酒组织形态分类方法和改进的欧式距离等技术开发一个基于葡萄叶片数字图像的葡萄品种自动识别软件。共测试17种酿酒和3种野生葡萄,识别率达87%。该方法具有自动化、识别速度快、花费低、省时省力等特点,可应用于葡萄病虫害的识别、杂交新品种双亲的判定等。     

Automatic classification of legumes using leaf vein image features

In this paper, a procedure for segmenting and classifying scanned legume leaves based only on the analysis of their veins is proposed (leaf shape, size, texture and color are discarded). Three legume species are studied, namely soybean, red and white beans. The leaf images are acquired using a standard scanner. The segmentation is performed using the unconstrained hit-or-miss transform and adaptive thresholding. Several morphological features are computed on the segmented venation, and classified using four alternative classifiers, namely support vector machines (linear and Gaussian kernels), penalized discriminant analysis and random forests. The performance is compared to the one obtained with cleared leaves images, which require a more expensive, time consuming and delicate procedure of acquisition. The results are encouraging, showing that the proposed approach is an effective and more economic alternative solution which outperforms the manual expert's recognition.     

一种用于农作物叶部病害图像识别的双权重协同表示分类方法

农作物病害是我国主要的农业灾害之一,准确识别病害类型是防治农作物病害的关键。因此,首先采集了小麦、玉米、花生、棉花4种农作物的22种常见叶部病害的441张图像;然后,在对每张病害图像中的叶片和病斑进行分割的基础上,分别提取了描述农作物种类的叶片特征参数和描述病害类型的病斑特征参数;其次,将这两类特征参数组合并作归一化处理,得到病害图像的数据特征向量;再次,采用所有病害图像的数据特征向量,构建了一个农作物叶部病害数据集;最后,在同时考虑数据特征重要性和数据空间局部性的基础上,提出了一种双权重协同表示分类(DWCRC)方法并将其用于农作物叶部病害识别。在农作物叶部病害数据集上的实验结果表明,提出的双权重协同表示分类方法在用于农作物叶部病害识别时具有较高的识别率。     

Optimising the application of the Hough transform for automatic feature extraction from geoscientific images

We have adapted the Hough transform to extract linear features successfully from geoscientific datasets. The Hough transform is used in an automatic technique, which makes use of a parameter space to describe features of interest in images. This method has been widely applied in machine vision for recognition of features in highly structured images. Geoscientific data is more demanding. Features of interest within scenes of natural environments exist on all scales, are often partially obscured and the images are usually noisy. Pre-processing of images before the HT is essential. Adaptations of the HT to cope with particular properties of geoscientific data include: optimising the dimensions of the discrete transform domain; using feature-modelling to cancel lines found; transforming multi-scale tiles of the original image and correcting amplitudes in the transformed domain to account for the position of features. These specific adaptations produce a method for automatic feature detection which requires the user to select only two parameters. Output of the procedure is rich in feature content and accurate, leaving a clean result for statistical analysis. This optimised HT is robust for natural scenes, coping in particular with short line-segments.     

Fractal dimension applied to plant identification

This article discusses methods to identify plants by analysing leaf complexity based on estimating their fractal dimension. Leaves were analyzed according to the complexity of their internal and external shapes. A computational program was developed to process, analyze and extract the features of leaf images, thereby allowing for automatic plant identification. Results are presented from two experiments, the first to identify plant species from the Brazilian Atlantic forest and Brazilian Cerrado scrublands, using fifty leaf samples from ten different species, and the second to identify four different species from genus Passiflora, using twenty leaf samples for each class. A comparison is made of two methods to estimate fractal dimension (box-counting and multiscale Minkowski). The results are discussed to determine the best approach to analyze shape complexity based on the performance of the technique, when estimating fractal dimension and identifying plants.     

基于图像处理技术的浮游生物自动分类研究

目前利用图象处理和模式识别技术进行浮游生物的自动分类与计数是一个研究热点,该文根据浮游蚤类图像的特点,设计了一种自动分类计数算法。该算法首先通过有效的分割手段将蚤类个体从背景区域中提取出来;然后计算与其生物学特征相关的形状和纹理特征参量;最后选择特征参量输入神经网络分类器对不同蚤类进行分类和计数。样本主要采集自胶州湾海域,图像通过zeiss显微图像采集系统获取。实验证明,由于研究目标明确,算法简单有效,取得了很好的分类计数效果,很大程度上减轻了人工分类计数的工作量。     

Use of IFS Codes for Learning 2D Isolated-Object Classification Systems

Automatic recognition of complex images is a hard and computationally expensive task, mainly because it is extremely difficult to capture in an automatic way and with a few features the necessary discriminant information. If such features were available, a proper learning system could be trained to distinguish images of different kinds of objects, starting from a set of labeled examples. In this paper we show that fractal features obtained from Iterated Function System encodings capture the kind of information that is needed by learning systems and, thus, allow the successful classification of 2-dimensional images of objects. We also present a fractal feature extraction algorithm and report the classification results obtained on two very different test-beds by applying Machine Learning techniques to sets of encoded images.     

基于卷积神经网络的苹果栽培品种识别

针对苹果栽培品种识别分类问题,提供一个包含多个苹果果树品种的叶片图像原始数据集,并且研究构建一种新的深度卷积神经网络分类模型,对其分类准确性、泛化性能和稳定性进行对比验证,以期对苹果栽培品种简便、快速、准确的识别分类提供理论依据和技术支持。以甘肃省平凉市静宁县果树果品研究所苹果良种苗木繁育基地作为实验基地,在其中选取14个苹果果树品种。每个品种选取10棵左右树龄、树势、长势都存在差异的果树,采摘100片左右成熟的、无机械损伤的叶片,然后拍摄叶片图像建立数据集,进而利用卷积神经网络训练识别分类模型。本文针对苹果栽培品种识别分类,提供一个包含14个苹果果树品种共计14394张叶片图像的原始数据集,并且设计实现基于卷积神经网络的识别分类模型。实验结果表明,该识别分类模型有较高的准确率,训练集训练精度可以达到99.88%,验证集验证精度为94.36%,独立测试集的测试精度为90.49%。本文的研究结果可以为现代苹果田间种植及科研试验等实际场景提供力所能及的帮助,为深度卷积神经网络技术在植物品种识别分类实际应用场景提供参考,丰富深度学习在农业上的应用。     

Utilizing venation features for efficient leaf image retrieval

Most Content-Based Image Retrieval systems use image features such as textures, colors, and shapes. However, in the case of a leaf image, it is not appropriate to rely on color or texture features only as such features are very similar in most leaves. In this paper, we propose a new and effective leaf image retrieval scheme. In this scheme, we first analyze leaf venation which we use for leaf categorization. We then extract and utilize leaf shape features to find similar leaves from the already categorized group in a leaf database. The venation of a leaf corresponds to the blood vessels in organisms. Leaf venations are represented using points selected by a curvature scale scope corner detection method on the venation image. The selected points are then categorized by calculating the density of feature points using a non-parametric estimation density. We show this technique’s effectiveness by performing several experiments on a prototype system.     

基于改进VGGNet模型的外来入侵植物叶片识别方法

针对自然界中不同种类植物的叶片可能存在类间差异小而导致一些边缘轮廓相似的本土植物和外来入侵植物叶片识别错误的问题,提出一种PF-VGGNet模型。常用的VGGNet模型在图像分类上表现优秀,采用顺次连接的结构,可以很好地提取图像的高级语义信息特征,但一些图像浅层的轮廓和纹理特征也对分类起到关键作用。PF-VGGNet模型可以将浅层轮廓和纹理特征与网络深层高级语义信息融合,实现对植物叶片的自动识别。实验结果表明,PF-VGGNet模型对比其它算法在自建的外来入侵植物叶片数据集上取得了较好的识别效果,在训练集和测试集上的准确率分别为99.89%和99.63%。PF-VGGNet可以有效降低因叶片边缘轮廓相近导致识别错误的问题,能够快速识别外来入侵植物叶片,为防治外来植物入侵提供支持。     

应用规则的肺结节识别系统

为了检测胸部CT图像中的肺结节,提出一种基于应用规则的自动识别肺结节的系统。在识别系统中通过自动阈值法和轮廓跟踪法分割肺实质;采用OTSU算法分割肺实质中的感兴趣区域,对感兴趣区域的特征进行提取;选择对肺结节和血管区别度较大的特征。根据选取的这些特征设定识别肺结节的规则来确定肺结节的候选区域。实验结果表明,该系统对直径1 cm以上的结节具有较好的识别性能。     

基于投影特征与结构特征的表格图像识别

通过分析表格的框线特征与结构特征,提出一种基于投影特征与结构特征的表格文本图像识别算法。该方法通过投影计算提取表格的框线特征,通过击中或击不中变换提取表格的结构特征,根据所提特征重要性的不同,设定分类判决阈值。实验结果表明,该方法能准确高效地区分表格文本图像与非表格文本图像,具有很强的实用性。     

Feature-based registration of histopathology images with different stains: An application for computerized follicular lymphoma prognosis

Follicular lymphoma (FL) is the second most common type of non-Hodgkin’s lymphoma. Manual histological grading of FL is subject to remarkable inter- and intra-reader variations. A promising approach to grading is the development of a computer-assisted system that improves consistency and precision. Correlating information from adjacent slides with different stain types requires establishing spatial correspondences between the digitized section pair through a precise non-rigid image registration. However, the dissimilar appearances of the different stain types challenges existing registration methods.This study proposes a method for the automatic non-rigid registration of histological section images with different stain types. This method is based on matching high level features that are representative of small anatomical structures. This choice of feature provides a rich matching environment, but also results in a high mismatch probability. Matching confidence is increased by establishing local groups of coherent features through geometric reasoning. The proposed method is validated on a set of FL images representing different disease stages. Statistical analysis demonstrates that given a proper feature set the accuracy of automatic registration is comparable to manual registration.     

基于几何不变量的图像特征识别

图像的特征识别是图像处理和识别中的一个重要问题，几何不变量作为特征的特征值在很多领域已经得到了广泛的应用。实际中，普遍采用在仿射变换及射影变换下保持不变的仿射、射影不变量作为特征值。本文根据具体图像的特点，利用4类仿射和射影不变量构成特征的特征值空间，依据4步识别策略来识别图像中的特征点，从而完成识别任务。实验表明，这4类不变量能够较好地识别出实际图像中的特征。     

基于ENM-Gabor差分权重的人脸表情特征提取方法

针对人脸子区域对表情识别分类的重要程度不同,提出一种基于Gabor小波特征和ENM(Eye,Nose,Mouth)差分权重的表情特征提取方法。通过对人脸眼睛、鼻子、嘴巴三个区域进行特征提取并自适应加以权重,有效地区分了不同区域对识别表情的重要程度。对预处理后的表情图像提取ENM区域Gabor特征;将表情图像与中性图像作差值计算得到ENM差分权重;将ENM-Gabor特征结合差分权重得到最终的表情特征并用BP神经网络进行分类。与其他方法在JAFFE表情库上进行对比实验,实验结果表明,该方法相比于传统Gabor特征提取有了明显的提高,且平均识别率达到99.3%。     

基于深度学习的盲文识别方法

提出一种基于深度学习的盲文点字识别方法,利用深度模型--堆叠去噪自动编码器（Stack Denoising AutoEncoder,SDAE）解决盲文识别中特征的自动提取与降维等问题。在构建深度模型过程中,采用非监督贪婪逐层训练算法（Greedy Layer Wise Unsupervised Learning Algorithm）初始化网络权重,使用反向传播算法优化网络参数。利用SDAE自动学习盲文点字图片特征,使用Softmax分类器进行识别。实验结果表明,本文所提方法较之传统方法,可以有效解决样本特征的自动学习与特征降维等问题,操作更为简易,并能获得满意的识别结果。     

基于特征加权的自动图像分类方法

低层特征的选择与提取是自动图像分类的基础,一方面,所选择的图像特征应能代表各种不同的图像属性,利于不同类别图像之间的区分;另一方面,为了提高后续模型的计算效率,需要减少噪声特征、冗余特征.提出了一种基于特征加权的自动图像分类方法.该方法根据图像低层特征分布的离散程度来衡量特征相对于类别的重要性,增加相关度高的特征的权重,降低相关度低的特征权重,从而避免后续模型被弱相关或不相关的特征所支配.所提的特征加权算法主要考察的是特征相对某个具体类别的重要程度,可以为每个类别选择出适合自身的特征权重.然后,将加权特征嵌入到支持向量机算法中用于自动图像分类,在Corel图像数据集上的实验结果表明,基于特征加权的自动图像分类算法可以有效地提高图像分类的准确性.     

一种基于粗糙集理论的票据图像识别分类方法

本文根据票据图像的版面特点,选取表格线、背景信息及纹理信息作为主要特征,采用粗糙集理论进行特征属性约简、规则提取及规则约简。并提出了一种基于多规则集的综合决策分类方法,巧妙的利用其分类及拒识的特点,有效地实现了票据图像分类。