Machine Vision Analysis Research Articles

3D Stereo Rendering Using FPGA

Stereo rendering presents a virtual 3D scene from two slightly different vantage points. It is of great importance in the field of machine vision, robotics and image analysis. This paper proposes a stereo vision system that is realized in a single field programmable gate array (FPGA). Calculations of the stereo pairs are made by using two-center projection (off-axis) method. The first red resultant image is for left eye while the second blue one is for right eye; the 3D illusion is produced when looking to them using anaglyph. This computer graphic hardware system is implemented using Spartan3E XC3S500E FPGA kit. The execution time for the proposal is 1266 faster than OpenGL time with maximum operating frequency of 35.417 MHz, while the max occupation area reaches 84%. Keywords: Computer Graphic; Stereoscopic; anaglyph; FPGA; two-center projection ;Off-axis Method; stereo pairs. DOI : 10.7176/CEIS/10-3-04 Publication date : April 30 th 2019

Development of woody plants phenotyping sys tem with the help of machine vision and spectral analysis algorithms

Phenotyping is a modern technology that allows registering and analyzing data about living systems phenotypes in automatic mode. In recent years a high-efficiency higher plants phenotyping with the aim of deep study of the physiological and biochemical bases of their functioning, as well as selection of new high-yielding and stress-resistant varieties has acquired significant development. It is based on adaptation and introduction of modern information approaches such as algorithms of «computer vision» and allows receiving detailed information about plants phenomes at various organization levels. One of the unreached research sections in this field is tree plant cuttings phenotyping, which represents a great practical interest. In this work the system of phenotyping of green stem cuttings of woody plants Thuja occidentalis L. (Smaragd), Juniperus scopulorum Sarg. (Blue Arrow), Picea Abies L., H. Karst. was developed using machine vision algorithms and spectral analysis. A modular phenomics complex was created. It consist of phenomics box, plant cultivation system, lighting and watering systems, as well as system of registration and processing of RGB-images, including software. This complex has been tested in the laboratory and in field conditions. The initial testing of the phenomics complex was carried out. It showed that the data of changes in the pixel distribution of RGB-images by wavelengths can be the basis for estimating the size and physiological state of stem cuttings of woody plants. It was shown that the shift of the spectral curve of the reflected light into the long-wave area under the influence of water deficit allows to non-invasive and statistically reliably register the death of some cells. At the same time, statistically reliable distinctions have not been revealed at testing of influence of phytohormones (auxins and brassinosteroids) on rooting of stem cuttings of coniferous species. The obtained results have a fundamental practical significance and can be used in the research of physiological processes in higher plants, ornamental crop production and forestry.

Automatic detection of ruminant cows’ mouth area during rumination based on machine vision and video analysis technology

Automatic detection of ruminant cows’ mouth area during rumination based on machine vision and video analysis technology

Machine vision analysis on abnormal respiratory conditions of mice inhaling particles containing cadmium

Inhalable environmental toxicants can induce pulmonary malfunction resulting abnormal respiratory conditions. The traditional methods currently available to detect the respiratory condition of animals rely on differential pressure transducers and signal amplifiers. In comparison, current machine vision application requires little hardware. But it is unsuitable for respiratory condition tests of experimental animals reflecting respiratory toxicities of inhalable pollutants. In this study, we establish a new automatic method of machine vision analysis using a model that has mice inhaling aqueous aerosol with different concentrations of CdCl2 (0, 1, 3, 5 mM 2 h/day) for 7 days as simulant occupational exposure of inhalable Cd and analyze respiratory conditions such as respiratory rate, rhythm index, drive index and exchange index. Additionally, the models with different degrees of lung damage in mice are further tested and verified by the concentrations of cadmium accumulated in the lungs and the analyses on pulmonary porosity, fibrosis and inflammation. Machine vision analysis can identify the abnormal respiratory conditions of mice. Respiratory rate and rhythm index increase after exposure to cadmium. In the individuals with mild lung damage, respiratory drive index and exchange index in treatment group are higher than that in the control group, and in individuals with severe lung damage, these indices are similar to that of the control group. These abnormal respiratory conditions related to variable lung damage in mice demonstrate that the respiration is synchronously influenced by inhalable Cd and respiratory compensation according to normal physiological regulation, suggesting the present method is effective.

Digital Foveation: An Energy-Aware Machine Vision Framework

In machine vision applications, imaging systems and analysis algorithms are generally interdependent and energy intensive. We describe a machine vision energy minimization framework in which imaging hardware and vision algorithms are co-designed and tightly integrated. Digital foveation is inspired by the human vision system, which uses a spatially varying sensing architecture to generate oculomotory feedback and capture a series of high-resolution images using the densely sampling fovea. A multiround process with bidirectional information flow between camera hardware and analysis software optimizes energy consumption while preserving accuracy. By using existing hardware mechanisms, namely, row / column skipping, random access via readout circuitry, and frame preservation, digital foveation adapts to the chosen analysis algorithm. It aims to transmit and process only the necessary parts of the scene under consideration. This framework is general across a wide range of embedded machine vision applications and enables large improvements in energy efficiency. When evaluated for an embedded license plate recognition vision application, it reduces system energy consumption by 81.3% with at most 0.65% reduction in accuracy.

Live-cell phenotypic-biomarker microfluidic assay for the risk stratification of cancer patients via machine learning.

The risk stratification of prostate cancer and breast cancer tumours from patients relies on histopathology, selective genomic testing, or on other methods employing fixed formalin tissue samples. However, static biomarker measurements from bulk fixed-tissue samples provide limited accuracy and actionability. Here, we report the development of a live-primary-cell phenotypic-biomarker assay with single-cell resolution, and its validation with prostate cancer and breast cancer tissue samples for the prediction of post-surgical adverse pathology. The assay includes a collagen-I/fibronectin extracellular-matrix formulation, dynamic live-cell biomarkers, a microfluidic device, machine-vision analysis and machine-learning algorithms, and generates predictive scores of adverse pathology at the time of surgery. Predictive scores for the risk stratification of 59 prostate cancer patients and 47 breast cancer patients, with values for area under the curve in receiver-operating-characteristic curves surpassing 80%, support the validation of the assay and its potential clinical applicability for the risk stratification of cancer patients.

Potato quality grading based on machine vision and 3D shape analysis

Machine vision is a non-destructive grading technology and cost-effective method with high accuracy that can be used to predict length, width, and mass, as well as defects of both interior and exterior of a sample by employing different cameras, such as color, multispectral, or hyperspectral cameras. To obtain certain data, which relates to sample quality in the 3D space (thickness, volume, and surface gradient distribution) and mass prediction, a novel method was developed and the obtained appearance quality was graded utilizing a new image processing algorithm for depth images. In this study, we recorded the depth images of 110 potatoes using a depth camera, including samples with uniform shapes or with deformations (e.g., bumps, bent shape, and divots). Length, width, thickness, and volume were calculated respectively, and used as key factors for detecting potato deformity, such as bent shape, bumps, and hollow. Experimental results indicate that mass prediction based on a volume model for both normal and deformed potato samples showed high accuracy, thus 90% of the samples were graded for the correct size group using the volume model. In addition, the appearance quality grading reached 88% of a correct percentage for bent shape, bump, and hollow defect detection by combining the surface data in 2D and 3D space. In addition, a potato virtual reality model rebuilding algorithm was developed for sample quality tracing and rechecking based on 3D shape and color images. This model redisplays the potato color and 3D shape data in multi-views and supports 360-degree rotation in both horizontal and vertical directions to simulate the in-hand examination experience. This depth image processing is an effective potential method for future non-destructive post-harvesting grading, especially for products where size, shape, and surface condition are important factors.

Clasificador de imágenes de frutas basado en inteligencia artificial

Las aplicaciones de visión artificial y el análisis de imágenes son capaces de resolver varios problemas del sector industrial, científico o de seguridad en la actualidad. La clasificación de imágenes es muy útil en la automatización de procesos en una empresa. Para realizar una tarea de clasificación de imágenes se requiere hacer la extracción de características que identifiquen a cada tipo de imagen como por ejemplo: color, forma, textura. En el presente trabajo se requiere implementan los algoritmos para la construcción de un clasificador de imágenes de frutas basado en la extracción de las características del color de las imágenes en determinadas regiones de interés. Para el desarrollo del clasificador de imágenes de frutas se utiliza la técnica de extracción del histograma a color en tres dimensiones y con la implementación de algoritmos de inteligencia artificial se efectúa la clasificación automática de imágenes. El conjunto de datos utilizado consiste en: cuatro clases de frutas con el número variable de imágenes por cada clase, luego se preparan las imágenes seleccionando las regiones de interés mediante técnicas de enmascaramiento y se las divide en dos grupos de datos: Los datos de entrenamiento y los datos de prueba. Luego de entrenado el clasificador, se realizan pruebas de clasificación para evaluar la eficacia del clasificador de imágenes de frutas. Esta metodología de construcción e implementación del clasificador se puede usar en varias aplicaciones según las clases de imágenes de objetos a analizar en casos similares y automatizar procesos de clasificación y reconocimiento de objetos.

Non-destructive crack detection of preserved eggs using a machine vision and multivariate analysis

Pidan or century egg, also known as preserved egg, is one of the most traditional and popular egg products in China. The crack detection of preserved eggshell is very important to guarantee its quality. In this study, we develop an image algorithm for preserved eggshell’s crack detection by using natural light and polarized image. Four features including crack length, crack state coefficient, maximum projection and angular point are extracted from the natural light image by morphology calculus algorithms. The support vector machines (SVM) model with radial basis kernel function is established using the four features with an accuracy of about 92%. The detection accuracy is improved to 94% by using a new characteristic parameter of crack length on polarization image. The Multi-information fusion analysis indicates the potential for cracks detection by a real-time synthesis imaging system.

A "Tuned" Mask Learnt Approach Based on Gravitational Search Algorithm.

Texture image classification is an important topic in many applications in machine vision and image analysis. Texture feature extracted from the original texture image by using “Tuned” mask is one of the simplest and most effective methods. However, hill climbing based training methods could not acquire the satisfying mask at a time; on the other hand, some commonly used evolutionary algorithms like genetic algorithm (GA) and particle swarm optimization (PSO) easily fall into the local optimum. A novel approach for texture image classification exemplified with recognition of residential area is detailed in the paper. In the proposed approach, “Tuned” mask is viewed as a constrained optimization problem and the optimal “Tuned” mask is acquired by maximizing the texture energy via a newly proposed gravitational search algorithm (GSA). The optimal “Tuned” mask is achieved through the convergence of GSA. The proposed approach has been, respectively, tested on some public texture and remote sensing images. The results are then compared with that of GA, PSO, honey-bee mating optimization (HBMO), and artificial immune algorithm (AIA). Moreover, feature extracted by Gabor wavelet is also utilized to make a further comparison. Experimental results show that the proposed method is robust and adaptive and exhibits better performance than other methods involved in the paper in terms of fitness value and classification accuracy.

Abstract 4347: A novel live cell diagnostic platform measuring phenotypic biomarkers using objective algorithmic analysis enables further risk stratification for intermediate-risk prostate cancer patients

Abstract Due to the inconsistencies of existing molecular, genomic, and pathophysiologic markers for patient risk stratification, effective prostate cancer diagnostics and treatment remains a challenge in clinical practice. Therefore, the development of a diagnostic platform that differentiates cancer patients who have clinically significant disease from those who have a low risk of progression is an important area of interest. In this study, we tested a diagnostic platform that combines a scalable microfluidic device, an automated live cell assay, and objective machine vision algorithms to measure phenotypic biomarkers [defined here as functional biophysical and molecular biomarkers], which evaluate both local growth and metastatic potential of prostate cancer. An analytical validation study was performed on fresh prostate cancer samples (n = 100) obtained at the time of radical prostatectomy (RP). The diagnostic platform enables: 1) growth of patient cells ex vivo on extra cellular matrix formulations supporting adhesion/survival for 72 hours 2) high-throughput imaging of multiple phenotypic biomarkers such as morphology, cytoskeleton dynamics, and protein subcellular localization & modification states and 3) objective quantification of biomarkers via machine vision analysis. Patient samples were imaged over a three hour period capturing live-cell biophysical biomarkers. After three hours cells were fixed and stained for molecular biomarkers. Machine vision technology was then utilized to analyze phenotypic biomarkers to yield specific metrics that quantified local tumor growth (Oncogenic Potential-OPs) and invasive potential of the tumor to other tissues (Metastatic Potential- MPs) that correlated with RP specimen pathologic findings. Analysis of quantified phenotypic biomarkers distinguished normal cells from cancer cells. The OP and MP metrics demonstrated statistical significance in distinguishing Gleason 6 (low-risk) from Gleason 7 (intermediate-risk) prostate cancer with 80% sensitivity and 80% specificity and concordance with relevant RP pathology findings. Specifically, OP and MP derived from defined phenotypic biomarker metrics, demonstrated the ability to differentiate Gleason 6 and 7 scores and correlated with, 1) seminal vesicle invasion, 2) positive RP surgical margins, 3) vascular invasion, and 4) lymph node involvement. This novel functional-live-cell diagnostic platform allows for the measurement of a biomarker panel that further stratifies patients to improve prostate cancer treatment, clinical decision-making, further risk stratification of intermediate prostate cancer populations, and potentially predict actionable pathological findings leading to improved treatment outcomes for prostate cancer patients. Citation Format: Michael S. Manak, Wendell R. Su, Andrew Min, Brad J. Hogan, Matthew J. Whitfield, Jonathan S. Varsanik, Delaney Berger, Mani Foroohar, Kimberly M. Rieger-Christ, Travis B. Sullivan, Naveen Kella, Ray Hernandez, Vladimir Mouraviev, Kevin B. Knopf, Hani H. Rashid, David M. Albala, Grannum R. Sant, Ashok C. Chander. A novel live cell diagnostic platform measuring phenotypic biomarkers using objective algorithmic analysis enables further risk stratification for intermediate-risk prostate cancer patients. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4347. doi:10.1158/1538-7445.AM2015-4347

Abstract LB-285: Computational pathology for predicting prostate cancer recurrence

Abstract Background: Conventional methods for predicting prostate cancer (PCa) aggressiveness, which rely heavily on the Gleason score of architectural pattern, remain imperfect. We have developed an approach that combines machine vision and machine learning analysis of routine H&E stained tumors to discriminate recurrent from non-recurrent patients after prostatectomy, based on architectural, color, and nuclear patterns. Methods: We used a set of H&E-stained TMAs containing quadruplicate cores from 176 recurrent cases of PCa and 154 non-recurrent controls, frequency matched on age, Gleason, pathological stage and race. Using software specifically adaptable for computational pathology with hierarchical object recognition (Definiens Developer XD®), we processed images of each tumor at two spatial scale levels: a larger scale for recognition of tissue compartments (epithelium, stroma, lumen and inflammation), and a smaller scale for recognition of normal and atypical nuclei in each of the compartments. Magnification of 20x was used for both levels. At the tissue level, we used multi-resolution segmentation to segment homogeneous superpixels, which were then classified using machine learning into epithelium, stroma. or ambiguous tissue classified as intermediate between epithelium and stroma. We used relative darkness in a spatial neighborhood, size, and shape criteria to segment typical and atypical nuclei. Relational features among objects within and between the tissue and nuclear levels were also created. The resultant feature library contained 884 base features and 3,551 total variables. We used several stages of iterative L1 penalized logistic regression and 5-fold cross-validation to reduce variables and obtain AUC estimates. Variable stability criteria applied across the folds further minimized model overfitting. Results: PSA, Gleason and CAPRA-s achieved AUCs < 0.60 for recurrence in this matched dataset. A fifth stage L1 model containing 18 histometric features had cross-validation AUC = 0.81 (95% CI: 0.76-0.85). A more permissive 74 feature model had AUC = 0.95 (95% CI: 0.93-0.97). Subgroup analysis restricted to Gleason 7 cases produced AUCs ranging from 0.79-0.93 with varying model sparseness. Including pre-surgical PSA, Gleason grade or CAPRA-s score did not change model performance. Notably, leading features were frequently derived from stromal or ambiguously stromal regions and involved nuclear shape or texture. Their independence from tumor grading reflected subject matching and lack of explicit accounting for nuclear or stromal characteristics in the Gleason criteria. Conclusion: A computational pathology approach, applied to discrimination of recurrent and non-recurrent PCa in routine H&E stains, shows considerable promise and appears capable of adding to the predictive power of Gleason and CAPRA scoring. We will conduct additional validation studies and extend this work to biopsy samples. Citation Format: Amit Sethi, Lingdao Sha, Ryan J. Deaton, Virgilia Macias, Andrew H. Beck, Peter H. Gann. Computational pathology for predicting prostate cancer recurrence. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-285. doi:10.1158/1538-7445.AM2015-LB-285

The obtainment and recognition of raw silk defects based on machine vision and image analysis

At present, the raw silk defects detection is traditional seriplane inspection that is greatly influenced by human factors and poorly repeatable. We introduce a method of machine vision and image analysis to intuitively detect the main kinds of raw silk defects which are loops and loose end in this paper. During the experiment, we develop an image acquisition system that includes a charge coupled device line scan sensor, a telecentric lens, a light source, and a raw silk winding device to capture the raw silk images continuously and steadily. After the image capture stage, the defect segmentation using thresholding and morphology operations, such as opening, hole fill up, and image subtraction, is carried out to extract the geometrical features accurately. To describe the defects under visual property, four geometrical features are extracted, which will be used as the input of BP neural network. A BP neural network is designed as a classifier to recognize the test samples. Experimented results indicate that the proposed method can successfully detect and recognize the main kinds of raw silk defects.

Characterization of Oats (Avena sativa L.) Cultivars Using Machine Vision

Machine vision or image analysis is an important tool in the study of morphology of any materials. This technique has been used successfully to differentiate the eleven oats cultivars based on morphological characters. The geometry of seeds was measured through image analyzer and the variation was observed and recorded. From the recorded data, the cluster analysis was carried out and it revealed that the cultivars could be grouped into two main clusters based on similarity in the measured parameters. Cultivar Sabzar, UPO 212, OL 9 and OL 88 formed one main cluster. The another main cluster includes cv. Kent, OS 6, UPO 94, HFO 114, OS 7, HJ 8 and JHO 822 with many sub clusters. Among the cultivars HJ 8 and JHO 822 has more similarity in all measured parameters than other cultivars. Thus morphological characterization through seed image analysis was found useful to discriminate the cultivars.

Quantifying blistering and browning properties of Mozzarella cheese. Part II: Cheese with different salt and moisture contents

Blistering and browning are two important properties when baking Mozzarella cheese on pizza. In this study, the blistering and browning of five Mozzarella cheeses with different salt and moisture contents were evaluated using machine vision and image analysis techniques. Cheese attributes, including meltability, free oil, elasticity and stretching properties, composition, water activity and microstructure, were also evaluated.All Mozzarella samples had similar browning appearances, due to their similar galactose contents. Mozzarella cheeses with higher salt concentration had relatively smaller fat globules and lower water activity, thus higher transition temperature, and elastic and stretching resistances which resulted in smaller blisters on the pizza. It was proposed that cheese with higher transition temperature had a shorter time to flow, and the elastic and stretching resistances of the melted cheese restrained the size of the blisters and impeded the growth of the blisters during heating.

Towards Sustainable Green Production: Exploring Automated Grading for Oil Palm Fresh Fruit Bunches (FFB) Using Machine Vision and Spectral Analysis

Over the last decade, Indonesian palm oil industry has become a leading producer of the world, and been able to generatenotable foreign export reserves. In spite of this, problems still persist in this industry, including low productivity due to mishandling of raw material in post-harvest operations. One of the prime causes of this is manual grading/sorting of fresh fruit bunches, which is prone to error and misjudgement, as well as subjectivity. High demand of oil palm establishes its high price in world market, which drives the industry to expand its plantation area to increase production. Ultimately, it compromise forests and agricultural land, resulting stagnation or decline in several food products. Alternatively, before expanding plantation extent, oil extraction productivity of existing plantation can be improved by carefully selecting appropriate FFBs for post-harvest processing through introduction of automation. The use of machine vision and spectral analysis has shown to assist productivity of agricultural processing industry. This study employs automation technology for FFB grading in oil palm mills, resulting in improved raw material quality, thereby increasing the oil extraction productivity, and simultaneously contributing to partly release the pressure of deforestation by maintaining green agricultural areas.

Two-Tier genetic programming: towards raw pixel-based image classification

Classifying images is of great importance in machine vision and image analysis applications such as object recognition and face detection. Conventional methods build classifiers based on certain types of image features instead of raw pixels because the dimensionality of raw inputs is often too large. Determining an optimal set of features for a particular task is usually the focus of conventional image classification methods. In this study we propose a Genetic Programming (GP) method by which raw images can be directly fed as the classification inputs. It is named as Two-Tier GP as every classifier evolved by it has two tiers, the other for computing features based on raw pixel input, one for making decisions. Relevant features are expected to be self-constructed by GP along the evolutionary process. This method is compared with feature based image classification by GP and another GP method which also aims to automatically extract image features. Four different classification tasks are used in the comparison, and the results show that the highest accuracies are achieved by Two-Tier GP. Further analysis on the evolved solutions reveals that there are genuine features formulated by the evolved solutions which can classify target images accurately.

Global-Based Salient Region Detection for Microscopic Image

The salient region detection has been a very important study in machine vision and image analysis. Reliable estimation of visual saliency allows appropriate processing of images without prior knowledge of their contents, and thus remains an important step in many computer vision tasks including image segmentation, object recognition, adaptive compression and image retrieval. This paper presents a global-based contrast region detection method. The color information and the relevance of spatial location were taken into account. Experimental results show that the proposed method compared with the existed methods, our method yielded better detection effect, more precise and low complexity, at the same time, the method was more applicable for salient region detection of microscopic image.

Mimetic finite difference methods in image processing

We introduce the use of mimetic methods to the imaging community, for the solution of the initial-value problems ubiquitous in the machine vision and image processing and analysis fields. PDE-based image processing and analysis techniques comprise a host of applications such as noise removal and restoration, deblurring and enhancement, segmentation, edge detection, inpainting, registration, motion analysis, etc. Because of their favorable stability and efficiency properties, semi-implicit finite difference and finite element schemes have been the methods of choice (in that order of preference). We propose a new approach for the numerical solution of these problems based on mimetic methods. The mimetic discretization scheme preserves the continuum properties of the mathematical operators often encountered in the image processing and analysis equations. This is the main contributing factor to the improved performance of the mimetic method approach, as compared to both of the aforementioned popular numerical solution techniques. To assess the performance of the proposed approach, we employ the Catte-Lions-Morel-Coll model to restore noisy images, by solving the PDE with the three numerical solution schemes. For all of the benchmark images employed in our experiments, and for every level of noise applied, we observe that the best image restored by using the mimetic method is closer to the noise-free image than the best images restored by the other two methods tested. These results motivate further studies of the application of the mimetic methods to other imaging problems. Mathematical subject classification: Primary: 68U10; Secondary: 65L12.

Using Image Analysis to Predict the Weight of Alaskan Salmon of Different Species

After harvesting, salmon is sorted by species, size, and quality. This is generally manually done by operators. Automation would bring repeatability, objectivity, and record-keeping capabilities to these tasks. Machine vision (MV) and image analysis have been used in sorting many agricultural products. Four salmon species were tested: pink (Oncorhynchus gorbuscha), red (Oncorhynchus nerka), silver (Oncorhynchus kisutch), and chum (Oncorhynchus keta). A total of 60 whole fish from each species were first weighed, then placed in a light box to take their picture. Weight compared with view area as well as length and width correlations were developed. In addition the effect of "hump" development (see text) of pink salmon on this correlation was investigated. It was possible to predict the weight of a salmon by view area, regardless of species, and regardless of the development of a hump for pinks. Within pink salmon there was a small but insignificant difference between predictive equations for the weight of "regular" fish and "humpy" fish. Machine vision can accurately predict the weight of whole salmon for sorting.