An active contour model (ACM) based on grayscale morphology fitting energy for fast image segmentation in the presence of intensity inhomogeneity is proposed. The core idea of grayscale morphology fitting energy is using the grayscale erosion and dilation operations to fit the image intensities on the two sides of contours. By extracting local intensity information using morphological operators, the proposed model can effectively segment images with intensity inhomogeneity, and the computational cost is low because the grayscale morphology fitting functions do not need to be updated during the process of curve evolution. Experiments on synthetic and real images have shown that the proposed model can achieve accurate segmentation. In addition, it is more robust to the choice of initial contour and has a higher segmentation efficiency compared to traditional local fitting-based ACMs.
The main goals of this study are to investigate the potential of absorption coefficient for the prediction of water contents in ‘Yuanhuang’ pear and analyze the relationship between the shelf-life and bulk optical properties in the range of 900-1050 nm. An automated integrating sphere (AIS) system was used to measure the total reflectance, total transmittance of pear flesh tissues in visible-Near infrared (Vis-NIR) range. These two measurements were used to estimate the absorption coefficient <i>μ<sub>a</sub></i> and reduced scattering coefficient <i>μ'<sub>s</sub></i> of pear samples by using an inverse adding doubling (IAD) light propagation model. The detection accuracy of the AIS system was verified by using both liquid (Intralipid-20% as scatterer) and solid phantom (TiO<sub>2</sub> as scatterer, carbon black as absorber). The relative error of measurement of <i>μ'<sub>s</sub></i> of liquid phantom with four different concentration (0.5%,1%,1.5%,2%) at 632.8 nm, 751 nm, 833 nm are less than 10% except for 2% concentration at 833 nm, and the relative error of measurement <i>μ<sub>a</sub></i> and <i>μ'<sub>s</sub></i> of solid phantom at 525.4 nm, 632.1 nm, 710.3 nm and 780.1 nm are less than 5% except for the <i>μ<sub>a</sub></i> at 525.4 nm. A total of 140 samples were used to conduct the moisture measurement, and drying method was used. Predictive models for moisture content from <i>μ<sub>a</sub></i> data were constructed using partial least squares regression (PLSR). The coefficient of correlation of calibration set (R<sub>c</sub>) and validation set (R<sub>p</sub>) were 0.50 and 0.45 respectively. The relationship between the shelf-life and optical properties was analyzed by dividing pear samples into three categories according to the actual shelf-life, and calculating classification accuracy by using actual and calculated shelf-life grade.
Korla fragrant pears are small oval pears characterized by light green skin, crisp texture, and a pleasant perfume for
which they are named. Anatomically, the calyx of a fragrant pear may be either persistent or deciduous; the deciduouscalyx
fruits are considered more desirable due to taste and texture attributes. Chinese packaging standards require that
packed cases of fragrant pears contain 5% or less of the persistent-calyx type. Near-infrared hyperspectral imaging was
investigated as a potential means for automated sorting of pears according to calyx type. Hyperspectral images spanning
the 992-1681 nm region were acquired using an EMCCD-based laboratory line-scan imaging system. Analysis of the
hyperspectral images was performed to select wavebands useful for identifying persistent-calyx fruits and for identifying
deciduous-calyx fruits. Based on the selected wavebands, an image-processing algorithm was developed that targets
automated classification of Korla fragrant pears into the two categories for packaging purposes.
The classification of walnuts shell and meat has a potential application in industry walnuts processing. A dark-field illumination method is proposed for the inspection of walnuts. Experiments show that the dark-field illuminated images of walnut shell and meat have distinct text patterns due to the differences in the light transmittance property of each. A number of rotation invariant feature analysis methods are used to characterize and discriminate the unique texture patterns. These methods include local binary pattern operator, wavelet analysis, circular Gabor filters, circularly symmetric gray level co-occurrence matrix and the histogram-related features. A recursive feature elimination method (SVM-RFE), is used to remove uncorrelated and redundant features and to train the SVM classifier at the same time. Experiments show that, by using only the top six ranked features, an average classification accuracy of 99.2% can be achieved.
Applications of machine vision in automated inspection and sorting of fruits have been widely studied by scientists and. Preprocess of the fruit image is needed when it contain much noise. There are many methods for image denoise in literatures and can acquire some nice results, but which will be selected from these methods is a trouble problem. In this research, total variation (TV) and shock filter with diffusion function were introduced, and together with other 6 common used denoise method s for different type noise type were tested. The result demonstrated that when the noise type was Gaussian or random, and SNR of original image was over 8,TV method can achieve the best resume result, when the SNR of original image was under 8, Winner filter can get the best resume result; when the noise type was salt pepper, median filter can achieve the best resume result
A new method along with shape descriptor using support vector machine for classify fruit shape is developed, the image is
first subjected to a normalization process using its regular moments to obtain scale and translation invariance, the rotation invariant
Zernike features are then extracted from the scale and translation normalized images and the numbers of features are decided by
primary component analysis (PCA), at last, these features are input to support vector machine (SVM) classifier and are compared to
different classifiers. This method using support vector machine as classifier performs better than traditional approaches that is
verified by some experiments.
The Gradient Vector Flow (GVF) snake was used for color fruit shape detection, which is proposed by Chengxiang
Xu, this snake has two well properties than traditional snake: large capture range and its ability to move into
boundary concavities. Indicator and morphological operation before applying GVF snake firstly preprocess the color
fruit image. In our experiments, we compared the detection result of this approach to traditional snake and traditional
edge operators and it is obvious that the performance of this approach is better; the boundaries detected by GVF
snake are thin and smooth, which are very important for fruit size detection and shape classification.
A machine vision system for egg weight detection was developed. Egg image was grabbed by a CCD camera and a frame grabber. An indicator composed of R, G, B intensity was used for image segmentation. A series of algorithms were developed to evaluate egg's vertical diameter, maximal horizontal diameter, upper horizontal diameter and nether horizontal diameter. Based on extracted four size features of vertical and maximal/upper/nether horizontal diameter, a regression model between egg's weight and its size was established using SAS, which was used to detect egg's weight. The experiment results indicated that, for egg weight detection on the machine vision system, the correlative coefficient of the regression model was 0.9781 and the absolute error was no more than ±3 g, which would be lower work load on human graders and an increased flexibility in the egg quality control process in egg's industrialization.
A machine vision system for real-time fruit quality inspection was developed. The system consists of a chamber,
a laser projector, a TMS-7DSP CCD camera (PULNIX Inc.), and a computer. A Meteor-II/MC frame grabber
(Matrox Graphics Inc.) was inserted into the slot of the computer to grab fruit images. The laser projector and the
camera were mounted at the ceiling of the chamber. An apple was put in the chamber, the spot of the laser
projector was projected on the surface of the fruit, and an image was grabbed. 2 breed of apples was test, Each
apple was imaged twice, one was imaged for the normal surface, and the other for the defect. The red component
of the images was used to get the feature of the defect and the sound surface of the fruits. The average value,
STD value and comentropy Value of red component of the laser scatter image were analyzed. The Standard
Deviation value of red component of normal is more suitable to separate the defect surface from sound surface
for the ShuijinFuji apples, but for bintang apples, there is more work need to do to separate the different surface
with laser scatter image.
In this research, a new algorithm for fruit shape classification was proposed. The level set representations according to signed distance transforms were used, which are a simple, robust, rich and efficient way to represent shapes. Based on these representations, the rigid transform was adopted to align shapes within the same class, and the simplest possible criterion, the sum of square differences was considered. After align procedure, the average shape representations can easily be derived and shape classification was performed by the nearest neighbor method. Promising results were obtained on experiments showing the efficiency and accurate of our algorithm.
Key words: machine vision, shape classification, fruit sorting, level set
A real time machine vision system for fruit quality inspection was developed, which consists of rollers, an encoder, a lighting chamber, a TMS-7DSP CCD camera (PULNIX Inc.), a computer (P4 1.8G, 128M) and a set of grading controller. An image was binary, and the edge was detected with line-scanned based digit image description, and the MER was applied to detected size of the fruit, but failed. The reason for the result was that the test point with MER was different from which was done with vernier caliper. An improved method was developed, which was called as software vernier caliper. A line between weight O of the fruit and a point A on the edge was drawn, and then the crossed point between line OA and the edge was calculated, which was noted as B, a point C between AB was selected, and the point D on the other side was searched by a way to make CD was vertical to AB, by move the point C between point A and B, A new point D was searched. The maximum length of CD was recorded as an extremum value. By move point A from start to the half point on the edge, a serial of CD was gotten. 80 navel oranges were tested, the maximum error of the diameter was less than 1mm.
A real time machine vision system for fruit quality inspection was developed, which consists of rollers, an encoder, a lighting chamber, a TMS-7DSP CCD camera (PULNIX Inc.), a computer (P4 1.8G, 128M) and a set of grading controller. The system was made for size detecting of fruit, and then sorting fruits into 3 groups by the skin color: red group, yellow group, and green group which was immaturity. Color model for segmenting fruits from background and classing fruits into different groups was discussed. RGB color model was used to segment fruits from background, an equation of red component and blue component was used to segment the figure of relationship between red and blue component into two zones, which represent background and a fruit respectively. And then HIS color model was introduced to class fruits into three groups, Hue component was used as the optimum feature for this objective because that there were less overlap on this component of the three groups.100 navel orange was used to class by their skin color, total error was 2.1%.
Shape is one of the major concerns and which is still a difficult problem in automated inspection and sorting of fruits. In this research, we proposed the multi-scale energy distribution (MSED) for object shape description, the relationship between objects shape and its boundary energy distribution at multi-scale was explored for shape extraction. MSED offers not only the mainly energy which represent primary shape information at the lower scales, but also subordinate energy which represent local shape information at higher differential scales. Thus, it provides a natural tool for multi resolution representation and can be used as a feature for shape classification. We addressed the three main processing steps in the MSED-based shape classification. They are namely, 1) image preprocessing and citrus shape extraction, 2) shape resample and shape feature normalization, 3) energy decomposition by wavelet and classification by BP neural network. Hereinto, shape resample is resample 256 boundary pixel from a curve which is approximated original boundary by using cubic spline in order to get uniform raw data. A probability function was defined and an effective method to select a start point was given through maximal expectation, which overcame the inconvenience of traditional methods in order to have a property of rotation invariants. The experiment result is relatively well normal citrus and serious abnormality, with a classification rate superior to 91.2%. The global correct classification rate is 89.77%, and our method is more effective than traditional method. The global result can meet the request of fruit grading.
The laws of gray distortion of machine vision system were discussed, and a method for gray calibration was presented. Five standard templates with unanimous gray value were used as the research objects. The average gray values of X direction and Y direction of the standard template images were obtained according to row and column. The gray distortion models were developed with moving average model of two image pixels. The models of five standard templates were developed separately, and the correlation coefficients of each model were above 0.96. The parameters of the gray distortion model were independent to the templates themselves. The gray calibration models of row and column were developed based on the gray distortion models separately, and the image gray values of other templates were proportion to the true value after gray calibration with the gray calibration models. The test verified the method.
A real time machine vision system for fruit size inspection was developed, which solved the problems such as fast processing the large amount of image information, improving system performance for real time dynamic image capture and processing capability, increasing precision of detection etc. For each fruit, four images were caught, and from which all the quality information of the whole surface were collected. Images were grabbed with a CCD camera (TMC-7DSP) and a frame grabber (Matrox Meteor II/MC), which is described in RGB space. The value of R/B was used as an index for image binary threshold after blurred image restoration. Median filter was used to denoise before edge detecting with Laplace Operator. A sphere fruit size-inspecting model was set up with a set of standard ball to calibrate the fruit size after the relative size of fruit, which was obtained with the method of partition edge point sets. The absolute error of the system was less than 1.1 mm and inspecting rate was over 31 fruits per second. That was this method can obtain fair inspecting speed, small absolute error, and filled the requirement of fruit automatic fruit sorting. But something is need to be paid attention, if shadow being in this vision system, it will arise big error when use partitions edge point, so it is needed to avoid the shadow.
A line-scanned based digit image description method was developed, where a digit image was scanned horizontally and the continuous pixel with similar information was described as a line segment, only the horizontal coordinate of the start pixel and the end were recorded to a node, and all the nodes on the same row were linked as a linked list called horizontal line list, all the horizontal line list of the image were stored in an array called image list by their vertical coordinate. The adjacent relationship between two vertical neighbor line segment was judged by the end coordinate of one line segment and the start of the others, in this way, all the adjacent horizontal line were move to a new array from image list to symbol an object, which was segmenting, where the operating of image filtering, object detecting, contour tracing was finished in one times. The method was applied to a realtime machine vision system (the computer is P4 1.8G, 128M RAM) for fruit quality inspection. The image resolution was 100 x 120, and the image of fruit on the image was about 40% to the whole. The processing rate was over 180 images per second.
The analysis of the vibration responses of a fruit is suggested to measure firmness non-destructively. A wooden ball excited the fruits and the response signals were captured using an accelerometer sensor. The method has been well studied and understood on ellipsoidal shaped fruit (watermelon). In this work, using the finite element simulations, the applicability of the method on watermelon was investigated. The firmness index is dependent on the mass, density, and natural frequency of the lowest spherical modes (under free boundary conditions). This developed index extends the firmness estimation for fruits or vegetables from a spherical to an ellipsoidal shape. The mode of Finite element analysis (FEA) of watermelon was generated based on measured geometry, and it can be served as a theoretical reference for predicting the modal characteristics as a function of design parameters such as material, geometrical, and physical properties. It was found that there were four types of mode shapes. The 1st one was first-type longitudinal mode, the 2nd one was the second-type longitudinal mode, the 3rd one was breathing mode or pure compression mode, and the fourth was flexural or torsional mode shape. As suggested in many references, the First-type spherical vibration mode or oblate-Prolate for watermelon is the lowest bending modes, it's most likely related to fruit firmness. Comparisons of finite element and experimental modal parameters show that both results were agreed in mode shape as well as natural frequencies. In order to measure the vibration signal of the mode, excitation and sensors should be placed on the watermelon surface far away from the nodal lines. The excitation and the response sensors should be in accordance with vibration directions. The correlations between the natural frequency and firmness was 0.856, natural frequency and Young's modulus was 0.800, and the natural frequency and stiffness factor (SF) was 0.862. The stiffness factor (SF) is adequate expression for the Modulus of Elastic (MOE), and adopted in the evaluation of their firmness.