Machine Vision Approach Research Articles

The Shannon Entropy Trend of a Fish System Estimated by a Machine Vision Approach Seems to Reflect the Molar Se:Hg Ratio of Its Feed.

The present study investigates the suitability of a machine vision-based method to detect deviations in the Shannon entropy (SE) of a European seabass (Dicentrarchus labrax) biological system fed with different selenium:mercury (Se:Hg) molar ratios. Four groups of fish were fed during 14 days with commercial feed (control) and with the same feed spiked with 0.5, 5 and 10 mg of MeHg per kg, giving Se:Hg molar ratios of 29.5 (control-C1); 6.6, 0.8 and 0.4 (C2, C3 and C4). The basal SE of C1 and C2 (Se:Hg > 1) tended to increase during the experimental period, while that of C3 and C4 (Se:Hg < 1) tended to decrease. In addition, the differences in the SE of the four systems in response to a stochastic event minus that of the respective basal states were less pronounced in the systems fed with Se:Hg molar ratios lower than one (C3 and C4). These results indicate that the SE may be a suitable indicator for the prediction of seafood safety and fish health (i.e., the Se:Hg molar ratio and not the Hg concentration alone) prior to the displaying of pathological symptoms. We hope that this work can serve as a first step for further investigations to confirm and validate the present results prior to their potential implementation in practical settings.

Cashews whole and splits classification using a novel machine vision approach

Cashew is a high-value and the third largest edible tree nut that is universally consumed as a snack or included in food preparations. Classification of whole and split cashews in industries is carried out manually by visual inspection and hand-picking, which is time-consuming, tedious, challenging, and laborious. A machine vision methodology of capturing cashew shadows and associated image processing ImageJ plugin to classify cashews into whole and splits were developed and tested. The developed classification algorithm was based on a novel idea of using surface grayscale-intensity-profile for split-up cashews and object shadows for split-down and whole cashews. Out of several features derived from grayscale-intensity-profile values, the “length of curve” best classified the split-up cashews from others. The challenge of classifying the whole and split-down cashews was addressed by a new “shadow to total-area ratio.” An accuracy of 100 % was achieved by the algorithm. A process scale-up by increasing the height and power of light source was also proposed. The promising results suggest that the developed algorithm can be coupled with a suitable hardware system to perform accurate classification of the whole and split cashews.

Machine vision based evaluation of impact of light emitting diodes (LEDs) on shoot regeneration and the effect of spectral quality on phenolic content and antioxidant capacity in Swertia chirata

The present study demonstrates the influence of LED irradiance of various wavelengths on shoot regeneration, biomass accumulation, photosynthetic pigment contents, and antioxidant potentials of Swertia chirata - a critically endangered medicinal plant. Mixed treatment of blue (BL) and red LEDs (RL) in equal proportion (1:1) significantly improved the shoot regeneration response. A machine vision system was developed to assess the shoot regeneration potential under different lighting treatments. Regenerated shoots exposed under BL:RL (1:1) exhibited higher biomass accumulation and canopy development compared to other lighting treatments. Improved canopy growth was evident from the increase in the area, major axis, minor axis, convex area, equivalent diameter and perimeter of regenerated shoot clusters. A higher correlation of dry weight (DW) was noted with the image feature, weighted density (WD) than the fresh weight (FW) in all the LED treated cultures. The significant correlation between DW and WD implies that the image feature WD can be adopted as a non-invasive approach for measuring biomass accumulation as well as detecting hyperhydricity. The developed machine vision approach provides a new direction in the evaluation of shoot organogenesis that displayed features including both shoot multiplication and canopy development. Chlorophyll and carotenoid contents of the regenerated shoots were found to be higher under BL:RL (1:1) than the other treatments. Supplementation of RL led to a reduction in the pigment contents. Spectral quality of lights also significantly influenced the accumulation of total phenolics, flavonoids and flavonols. Cultures exposed under BL exhibited the maximum accumulation of polyphenols. A similar effect of spectral quality was observed with the antioxidant capacity and reducing power potential of leaf extract. The findings demonstrate the ability of LEDs in inducing shoot regeneration as well as accumulation of phenolic antioxidants and suggest that the proportion of blue and red LEDs is an important factor in achieving the optimum response.

Noise estimation in cardiac x-ray imaging: a machine vision approach

We propose a method to automatically parameterize noise in cardiac x-ray image sequences. The aim was to provide context-sensitive imaging information for use in regulating dose control feedback systems that relates to the experience of human observers. The algorithm locates and measures noise contained in areas of approximately equal signal level. A single noise metric is derived from the dominant noise components based on their magnitude and spatial location in relation to clinically relevant structures. The output of the algorithm was compared to noise and clinical acceptability ratings from 28 observers viewing 40 different cardiac x-ray imaging sequences. Results show good agreement and that the algorithm has the potential to augment existing control strategies to deliver x-ray dose to the patient on an individual basis.

Automated status inspection of fastening bolts on freight trains using a machine vision approach

The status inspection and maintenance actions on the mechanical components of freight trains are significant determinants of railway safety. Fastening bolts, as a key component, are widely used to transmit, connect and fasten various parts of freight trains. Their absence can degrade the original structure and lead to accidents. Recently, machine vision approaches have been widely used to inspect the status of mechanical components, thereby reducing costs and avoiding traffic accidents. This paper proposes a visual inspection system that is based on the machine vision approach and can be used to automatically inspect the status of fastening bolts on freight trains. To detect the presence/absence of a fastening bolt in a complex background, a hierarchical detection framework consisting in fault area extraction and fastening bolt detection is proposed. In the first module, a gray projection method is used to divide the fault area that contains the target from the complex background. Subsequently, a fastening bolt detector is designed to verify the candidate image regions. Several gradient-orientation-based features and a classifier can be used to perform the detection task. Experimental results show that the combination of a gradient orientation co-occurrence matrix and a support vector machine has the best classification performance. The proposed inspection system has the advantages of good real-time performance and high inspection accuracy; it achieves an accuracy of 99.96% with a speed of nine frames per second.

Extended target tracking filter with intermittent observations

This study addresses the problem of tracking extended target with intermittent observations. Based on practical applications, two Bernoulli distributed random variables are employed to describe the intermittent phenomenon of the positional measurements and the measurements of target extent, respectively. First, a machine vision algorithm is developed to solve the target shape parameters. Then, four sub-filters are designed according to the received observations and the achieved target shape parameters. The output of the proposed tracking filer can be obtained by the weighted-confidence fusion of the sub-filters. Finally, the machine vision algorithm is evaluated by the virtual target images created in OpenGL (Open Graphics Library) and the real images of a moving ship. The performance of the designed tracking filter is compared with the traditional tracking filter. The experiment results show the effectiveness of the machine vision approach; also the Monte-Carlo runs demonstrate that the provided tracking filter outperforms the traditional one with respect to accuracy.

Analysis of spatters in laser welding with beam oscillation: A machine vision approach

Spatter formation is still an issue in welding, particularly in laser welding. This research takes up the topic and proposes a machine vision approach for spatter tracking in high speed image series. After a description of the multi hypothesis tracking method, in which a Kalman-filter is used for optimal object state estimation, the tracking is applied on high speed images from an experimental series on laser welding with beam oscillation using stainless steel. On the basis of its results three different spatter formation mechanisms could be identified and described. Those were spatter formation by material ablation, by periodic re-entry of the laser spot into the melt pool and by melt pool dynamics.

A two-camera machine vision approach to separating and identifying laboratory sprouted wheat kernels

A proof-of-concept for a two-camera machine vision (MV) to classify the laboratory sprouted wheat kernels into sound, sprout-damaged, or severely sprout-damaged classes was developed. The marker controlled watershed segmentation technique was tailored to disjoin the clustered kernels in the digital images. Images were captured for both dorsal and ventral sides of the kernels. Segmentation accuracy of at least 95% was achieved depending upon the sample size under some conditions. Sixteen features comprising of colour, texture, and shape and size were extracted from the images of dorsal and ventral sides of the test kernels. The alpha-amylase activities, a key enzyme found in the sprout-damaged wheat were measured analytically for each kernel to categorise it into one of the three classes. A neural network model was developed with the kernels' features as the inputs and the alpha-amylase activity as the output. The MV with the trained neural network classified the test wheat kernels into three classes with an accuracy of 72.8%. Some of the challenges associated with the system are discussed. Recommendations to improve the system accuracy and robustness, and to decrease the system cost are presented.

Tool breakage detection from 2D workpiece profile using vision method

In-process tool breakage monitoring can significantly save cost and prevent damages to machine tool. In this paper, a machine vision approach was employed to detect the tool fracture in commercial aluminium oxide ceramic cutting tool during turning of AISI 52100 hardened steel. The contour of the workpiece profile was captured with the aid of backlighting during turning using a high-resolution DSLR camera with a shutter speed of 1/4000 s. The surface profile of the workpiece was extracted to sub-pixel accuracy using the invariant moment method. The effect of fracture in ceramic cutting tools on the surface profile signature of the machined workpiece using autocorrelation was studied. Fracture in the aluminum oxide ceramic tool was found to cause the peaks of autocorrelation function of the workpiece profile to decrease rapidly as the lag distance increased. The envelope of the peaks of the autocorrelation function was observed to deviate significantly from one another at different workpiece angles when the tool has fractured due to the continuous fracture of ceramic cutting insert during machining.

Context sensitive cardiac x-ray imaging: a machine vision approach to x-ray dose control

Modern cardiac x-ray imaging systems regulate their radiation output based on the thickness of the patient to maintain an acceptable signal at the input of the x-ray detector. This approach does not account for the context of the examination or the content of the image displayed. We have developed a machine vision algorithm that detects iodine-filled blood vessels and fits an idealized vessel model with the key parameters of contrast, diameter, and linear attenuation coefficient. The spatio-temporal distribution of the linear attenuation coefficient samples, when appropriately arranged, can be described by a simple linear relationship, despite the complexity of scene information. The algorithm was tested on static anthropomorphic chest phantom images under different radiographic factors and 60 dynamic clinical image sequences. It was found to be robust and sensitive to changes in vessel contrast resulting from variations in system parameters. The machine vision algorithm has the potential of extracting real-time context sensitive information that may be used for augmenting existing dose control strategies.

What makes a Pollock Pollock: a machine vision approach

Jackson Pollock introduced a revolutionary artistic style of dripping paint on a horizontal canvas. Here we study Pollock's unique artistic style by using computational methods for characterising the low–level numerical differences between original Pollock drip paintings and drip paintings of other painters who attempted to mimic his signature drip painting style. Four thousands and twenty four numerical image content descriptors were extracted from each painting, and compared using weighted nearest neighbour classification such that the Fisher discriminant scores of the content descriptors were used as weights. In 93% of the cases, the computer analysis was able to differentiate between an original and a non–original Pollock drip painting. The most discriminative image content descriptors that were unique to the work of Pollock were the fractal features, but other numerical image content descriptors such as Zernike polynomials, Haralick textures, and Chebyshev statistics show substantial differences between original and non–original Pollock drip paintings. These experiments show that the uniqueness of Pollock's drip painting style is not reflected merely by fractals, but also by other numerical image content descriptors that reflect the visual content. The code and software used for the experiment is publicly available, and can be used to study the work of other artists.

Asynchronous vehicle pose correction using visual detection of ground features

The inherent noise associated with odometry manifests itself as errors in localization for autonomous vehicles. Visual odometry has been previously used in order to supplement classical vehicle odometry. However, visual odometry is limited in its ability to reduce errors in localization for large travel distances that entail the cumulative summing of individual frame-to-frame image errors. In this paper, a novel machine vision approach for tiled surfaces is proposed to address this problem. Tile edges in a laboratory environment are used to define a travel trajectory for the Quansar Qbot (autonomous vehicle) built on the iRobot iRoomba platform with a forward facing camera. Tile intersections are used to enable asynchronous error recovery for vehicle position and orientation. The proposed approach employs real-time image classification and is feasible for error mitigation for large travel distances.The average position error for an 8m travel distance using classical odometry was measured to be 0.28m. However, implementation of the proposed approach resulted in an error of 0.028m. The proposed approach therefore significantly reduces pose estimation error and could be used to supplement existing modalities such as GPS and Laser-based range sensors.

Effect of surface lay in the surface roughness evaluation using machine vision

This work explores the influence of orientation of surface lay pattern of the machined components, while quantifying the surface roughness using machine vision approach. The surface images are captured from milled low carbon steel specimens with different roughness values using a vision system with coaxial lighting arrangement at different angular orientations of the work pieces (0°, 30°, 45°, 60°, 90°, 120°, 135°, 150°, and 180°). The captured images are subjected to preprocessing in order to retain the frequency components that attribute to roughness using a Gaussian filter by adapting the filtering procedures specified in ISO 4288. Numerous image based parameters such as gray level average (Ga), gray level co-occurrence matrix based image quantification parameters namely contrast, correlation, energy or uniformity, maximum probability and differential box courting based fractal dimension are computed from the surface images captured at different angular positions of the work piece. The computed vision based parameters are compared and correlated with the roughness average (Ra) obtained using a stylus instrument and the results are analyzed. The results clearly indicated that it is important to consider the orientation of the work piece when the machine vision approach is used to quantify the surface texture parameters.

Progressive cutting tool wear detection from machined surface images using Voronoi tessellation method

Tool condition monitoring by machine vision approach has been gaining popularity day by day since it is a low cost and flexible method. In this paper, a tool condition monitoring technique by analysing turned surface images has been presented. The aim of this work is to apply an image texture analysis technique on turned surface images for quantitative assessment of cutting tool flank wear, progressively. A novel method by the concept of Voronoi tessellation has been applied in this study to analyse the surface texture of machined surface after the creation of Voronoi diagram. Two texture features, namely, number of polygons with zero cross moment and total void area of Voronoi diagram of machined surface images have been extracted. A correlation study between measured flank wear and extracted texture features has been done for depicting the tool flank wear. It has been found that number of polygons with zero cross moment has better linear relationship with tool flank wear than that of total void area.

Automated Raw Part Alignment by a Novel Machine Vision Approach

Large raw parts require a long time consuming process of alignment into the machine tool, before the machining process starts. Alignment process requires two steps: first, characterization of geometry, and second, alignment. Important skills are necessary, and, besides the time consumption of workforce and machine, the risk of getting into shortage of material is high. The paper presents a novel automated alignment solution based on 3D vision technology, with the aim of minimising the total raw part alignment process time. The potential of the solution to reduce the alignment process time in machine tools is demonstrated in milling pilot cases.

Machine Vision Approach for Automating Vegetation Detection on Railway Tracks

AbstractThe presence of vegetation on railway tracks (amongst other issues) threatens track safety and longevity. However, vegetation inspections in Sweden (and elsewhere in the world) are currently being carried out manually. Manually inspecting vegetation is very slow and time consuming. Maintaining an even quality standard is also very difficult. A machine vision-based approach is therefore proposed to emulate the visual abilities of the human inspector. Work aimed at detecting vegetation on railway tracks has been split into two main phases. The first phase is aimed at detecting vegetation on the tracks using appropriate image analysis techniques. The second phase is aimed at detecting the rails in the image to determine the cover of vegetation that is present between the rails as opposed to vegetation present outside the rails. Results achieved in the current work indicate that the machine vision approach has performed reasonably well in detecting the presence/absence of vegetation on railway tracks when compared with a human operator.

Simultaneous Imaging and Precision Alignment of Two mm Wave Antennas Based on Polarization-Selective Machine-Vision

In this paper, we present an optical imaging tool, the Overlay Imaging Aligner (OIA), developed to aid in the mechanical alignment of antenna components in the mm-wave and low-THz frequency regimes (50-500 GHz) where the millimeter and sub millimeter wavelengths pose significant alignment challenges. The OIA uses a polarization-selective machine-vision approach to generate two simultaneous and overlaid real-time digital images along a common axis; this allows for aligning two antenna components to within fractions of a wavelength in the mm-wave and THz frequency regimes. The overall concept, optical design, function, performance characteristics and application examples are presented. A quantitative assessment of the alignment accuracy achievable with the OIA at specific frequencies in the WR-2.2 band is made where the alignment achieved with the OIA is compared to an electrical alignment.

Precision Alignment of Two mm-Wave Antennas Using an Improved Optical Alignment Tool

In this paper, we present an optical imaging tool, the Overlay Imaging Aligner (OIA), developed to aid in the mechanical alignment of antenna components in the mm-wave and low-THz frequency regimes (50-500 GHz). In these regimes, the millimeter and sub-millimeter wavelengths pose significant challenges for alignment. The OIA uses a polarization-selective, machine-vision approach to generate two simultaneous and overlaid real-time digital images along a common axis. This allows for aligning two antenna components to within fractions of a wavelength in the mm-wave and THz frequency regimes. The overall concept, optical design, function, performance characteristics, and application examples are presented. Preliminary data at specific frequencies in the WR-2.2 band are presented that compare the alignment achieved with the OIA to an electrical alignment.

Comparison of particle size distribution of celestite mineral by machine vision ΣVolume approach and mechanical sieving

A reliable and accurate measurement of particle size and particle size distribution (PSD) is central to characterization of particulate minerals. Using mineral celestite (SrSO 4) as the test material, an inexpensive machine vision approach as an alternative to standard mechanical sieving was proposed and results were compared. The machine vision approach used a user-coded ImageJ plugin that processed the digital image in a sieveless manner and automated the PSD analysis. A new approach of employing sum of volumes (ΣVolume) as weighting factor was developed and utilized in the ASABE standard PSD analysis. The plugin also evaluated 22 significant dimensions characterizing samples and 21 PSD parameters. According to Folk and Ward's classification, the PSD of ball-milled celestite was “very finely skewed” and “leptokurtic”. The PSD of celestite followed a lognormal distribution, and the plot against particle size exhibited almost a linear trend for both machine vision and mechanical sieving methods. The cumulative undersize PSD characteristics of both methods matched closely when the width-based mechanical sieving results were transformed to lengths by applying the shape factor (width/length). Based on the study, this machine vision approach can be utilized for PSD analysis of particulate minerals and similar products.

Machine vision 3D skin texture analysis for detection of melanoma

PurposeThe purpose of this paper is to describe innovative machine vision methods that have been employed for the capture and analysis of 3D skin textures; and the resulting potential for assisting with identification of suspicious lesions in the detection of skin cancer.Design/methodology/approachA machine vision approach has been employed for analysis of 3D skin textures. This involves an innovative application of photometric stereo for the capture of the textures, and a range of methods for analysing and quantifying them, including statistical methods and neural networks.Findings3D skin texture has been identified as a useful indicator of skin cancer. It can be used to improve realism of virtual skin reconstructions in tele‐dermatology. 3D texture features can also be combined with 2D features to obtain a more robust classifier for improving diagnostic accuracy, thereby assisting with the long‐term goal of implementing computer‐aided diagnostics for skin cancer.Originality/valueThe device developed for capturing 3D skin textures is known as the “Skin Analyser”, and as far as the authors know it is unique in the world in being able to recover 3D textures from pigmented lesions in vivo. There currently exist numerous methods for analysing lesions, including manual inspection (using established heuristics commonly known as ABCD rules), dermoscopy and SIAoscopy. The ability to capture and analyse 3D lesion textures complements these existing techniques and forms a valuable additional indicator for assisting with the early detection of dangerous skin cancers such as melanoma.