CN210377552U

CN210377552U - Fruit is multiaspect image acquisition device for classification

Info

Publication number: CN210377552U
Application number: CN201921144321.3U
Authority: CN
Inventors: 杨小强
Original assignee: Chongqing Creation Vocational College
Current assignee: Chongqing Creation Vocational College
Priority date: 2019-07-22
Filing date: 2019-07-22
Publication date: 2020-04-21
Anticipated expiration: 2029-07-22

Abstract

The utility model belongs to the technical field of fruit classification, in particular to a multi-surface image acquisition device for fruit classification, which comprises a picture digital processing system for taking pictures and realizing fruit sample picture digitization, an upper PC (personal computer) and an image processing and information analysis system for analyzing surface quality and defect identification; the three-dimensional characteristics of the fruits, such as height, volume, color, shape and the like, are collected by the binocular camera, so that the size, variety and quality of the fruits can be reflected specifically, and the fruits are subjected to multi-azimuth and full-stereoscopic image locking by a binocular determination method, so that the sorting accuracy is improved; the appearance shape and the variety color of the fruits are subjected to multi-level grading identification to form a high-resolution grade evaluation system, so that the subsequent sorting machine can not only select sizes and quality, but also distinguish different varieties, and the fruits are sorted more accurately, quickly and efficiently.

Description

Fruit is multiaspect image acquisition device for classification

Technical Field

The utility model belongs to the technical field of fruit classification, concretely relates to fruit is categorised with multiaspect image acquisition device.

Background

Nowadays, the research on fruit sorting technology at home and abroad is mainly embodied in the development of fruit sorting algorithm based on machine vision. For example, a color camera integrated with a Charge Coupled Device (CCD) is used for acquiring surface pictures of fruits with different levels of quality, and a network model is put forward by multi-dimensional visual characteristic training, so that the fruits are quickly classified and sorted by combining with a developed mechanical removing device. The method is widely applied to sorting fruits such as pears, apples, oranges, tomatoes, kiwis and the like, the grading accuracy can be more than 90% according to different proposed network models, and the fastest sorting speed can reach 50 ms/fruit according to reports. The other method is to develop an algorithm aiming at the fruit image acquired by the spectral camera, and identify whether the surface of the fruit has regions with property differences according to different spectral wave bands reflected by the surface of the fruit, so as to realize the purpose of grading. Since this type of method is more directed to fruit surface defect detection, the first type of method is more suitable for rapid sorting development based on visual characteristics (e.g., size, shape, color) of the fruit surface. However, few studies on sorting techniques of fruits have been reported. Therefore, developing a corresponding classification algorithm and a sorting technique is a hot spot of research nowadays. And the classified fruits are formed into fruit commodities by using the designed packaging technology, so that the economic value of the fruits can be greatly exerted, and the method is a research hotspot of green industry.

The Youcai area is a high-yield region of fruits, is a production base of fruits such as mandarins, pears and the like in China, the processes of sorting and boxing a large amount of fruits are finished manually at present, the manual sorting mode is time-consuming and labor-consuming, the judgment standards of different people are inconsistent, the size and the color and the shape of each box of fruits are different, the sale price of the fruits is finally low, and the economic benefit of the fruit industry cannot be brought into full play. Currently, fruit sorting technology remains in the research stage. Departments including colleges and universities, and enterprises are engaged in the development of this technology. However, no efficient fruit sorting technology is reported, and no hardware pipeline market is available. Therefore, the research on the fruit sorting technology and the development of the packaging production line are significant for filling the blank of the Yongchuan district and the surrounding counties in the fruit sorting field.

SUMMERY OF THE UTILITY MODEL

The utility model provides a fruit is multiaspect image acquisition device for classification to solve the problem that proposes among the above-mentioned background art.

In order to achieve the above object, the utility model provides a following technical scheme: the utility model provides a fruit is multiaspect image acquisition device for classification, is including being used for taking the picture, realizes the digital picture digital processing system of fruit sample picture, host computer and with image processing and the information analysis system with analysis surface quality and defect identification, the picture digital processing system includes the camera and rather than complex board in a poor light, the camera passes through net twine and host computer signal connection, be equipped with in the host computer image processing and information analysis system, image processing and information analysis system include that characteristic extraction unit, grade and variety standard formulate unit and grade and variety evaluation unit, the characteristic extraction unit includes colour characteristic extraction unit and regional characteristic extraction unit.

Preferably, the camera is a binocular CCD area array industrial camera.

Preferably, the color feature extraction unit collects three color vectors of red, yellow, and blue.

Preferably, the region feature extraction unit acquires texture features and morphological features of the image.

Preferably, the upper PC is provided with a median filter and a binomial filter for eliminating noise and gaussian noise carried in the image.

Preferably, the picture digital processing system is arranged on the fruit conveying device.

Compared with the prior art, the beneficial effects of the utility model are that:

1. the three-dimensional characteristics of the fruits such as height, volume, color, shape and the like can be collected through the binocular camera to specifically reflect the size, variety and quality of the fruits, and the fruits are subjected to multi-azimuth and full-stereoscopic image locking through binocular determination, so that the sorting accuracy is improved.

2. The appearance shape and the variety color of the fruits are subjected to multi-level grading identification to form a high-resolution grade evaluation system, so that the subsequent sorting machine can not only select sizes and quality, but also distinguish different varieties, and the fruits are sorted more accurately, quickly and efficiently.

Drawings

The accompanying drawings are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention, and together with the description serve to explain the invention and not to limit the invention. In the drawings:

fig. 1 is a schematic structural view of the present invention;

fig. 2 is the state diagram of the image information acquired by the CCD lens of the present invention.

In the figure: 1. a picture digital processing system; 11. a camera; 12. a backlight plate; 2. an upper PC; 3. an image processing and information analysis system; 31. a feature extraction unit; 32. a color feature extraction unit; 33. a region feature extraction unit; 34. a grade and variety standard making unit; 35. grade and variety evaluation unit.

Detailed Description

The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments in the present invention, all other embodiments obtained by a person skilled in the art without creative work belong to the protection scope of the present invention.

Referring to fig. 1-2, the present invention provides the following technical solutions: the utility model provides a fruit is categorised with multiaspect image acquisition device, including the picture digital processing system 1 that is used for taking the picture, realize fruit sample picture digitization, upper PC 2 with the image processing and the information analysis system 3 of analysis surface quality and defect identification, picture digital processing system 1 includes camera 11 and rather than complex board 12 in a poor light, camera 11 passes through net twine and upper PC 2 signal connection, be equipped with image processing and information analysis system 3 in the upper PC 2, image processing and information analysis system 3 include characteristic extraction unit 31, grade and variety standard formulation unit 34 and grade and variety evaluation unit 35, characteristic extraction unit 31 includes color feature extraction unit 32 and regional characteristic extraction unit 33.

In this embodiment, the picture digitization processing system 1 is configured to take pictures by the camera 11, the backlight plate 12, and some auxiliary devices, such as light sources, etc., to digitize the pictures of the fruit samples and remove some interference factors, and the latter image processing and information analysis system 3 extracts features that can reflect the surface quality of the samples to analyze the surface quality and identify defects.

① Camera and lens selection

The field of view of the shot is 220mm x 300mm, the required detection accuracy is 1mm, then the lowest resolution of the camera should be chosen: (300/1) × (220/1) ═ 66000, in order to improve the precision and stability of the system, the area of the defect is taken to be more than 5 pixels, thus the camera that we choose is also more than 1100 × 1500, therefore choose CCD area array industrial camera as the image acquisition equipment, obtain clearly to detect the fruit image and digitize it, the area array camera and upper PC 2 use the net mouth to communicate in the system, can reduce the probability that the noise produces in the image transmission, choose the front lighting directly, the number of light sources and putting the way to be decided according to the experimental effect, the whole ambient light condition requires relatively invariable.

The basis of the selection of the industrial lens of the machine vision is matched with the system, so that the detection and measurement results are more accurate, and the stability of the system is ensured.

② image acquisition and processing

The image acquisition equipment selects a mode of fixing the background to continuously acquire images, the light brightness and the aperture and the focal length of the camera are required to be adjusted to a proper position and the focal length is locked when the image acquisition equipment is used, meanwhile, the external triggering photographing delay of the industrial camera is required to be matched with the speed adjustment of the conveyor belt, and the acquired images can be accurately transmitted to the upper PC 2 through the network port when the adjusted image acquisition equipment is used for processing.

Meanwhile, a median filter and a binomial filter in the spatial filtering technology are used for eliminating random noises such as noise, Gaussian noise and the like carried in the image, and an image enhancement technology is used for enhancing the blurred edge part of the image.

③ binocular coordinate calibration

The three-dimensional vision of the fruit detection system needs to establish a pixel coordinate system, an image coordinate system, a camera coordinate system, a base coordinate system and a manipulator tail end coordinate system, and the five coordinate systems need to calculate the conversion relation among the coordinate systems at the same time, so that a series of work such as measurement, positioning, sorting, grabbing and the like of a target object can be realized.

And the calibration of the monocular and binocular cameras is completed by utilizing the principle that the binocular stereoscopic vision acquires the disparity map and the three-dimensional coordinates, and data analysis is performed on the calibration result.

Assuming that projection points of a spatial point S (X, Y, Z) on imaging planes of the left camera and the right camera are respectively Sl (xl, yl) and Sr (xr, yr), a passing point O1 and an O2 are respectively taken as virtual imaging plane vertical lines, vertical points are sequentially Bl and Br, a passing point S is taken as a vertical line towards Bl Br, and an O1O2 is taken as a vertical line, and the vertical points are Sr and Bp. From the triangle similarity we can derive:

︱SBP︱/︱SOP︱＝︱SrBP︱/︱SrOP︱

︱SBP︱/︱SOP︱＝︱S1BP l︱/︱S1OP l︱

an | SOP | a, | APOP | F, | O1O2 | B, | blsll | D, | BrSr | R, | SrBp | C, then

The formula can be expressed as:

A-F/A＝C/R+C

A-F/A＝B-D+R+C/B+R+C

finally, the following can be obtained:

A＝(B/D-R)*F

wherein D-R is called binocular parallax, A is related to the distance between the optical centers of the left and right cameras, the focal length and the parallax of the left and right cameras, and B and F can be obtained by calibrating camera parameters, so that the binocular parallax is the only variable influencing depth information. The three-dimensional coordinates of the S-point can be found.

(2) Fruit grade classification system construction

① color extraction

The color image is composed of three color channel vectors, the skin color of the fruit determines the maturity and variety of the fruit, different fruit pictures are converted into data information, and corresponding fruit colors are extracted and analyzed through different color information effects collected by a camera.

② extraction and selection of textural features

Texture describes the metric relationship between local area pixels, and the pattern of gray scale in space obtained by specific morphological changes, therefore, describing texture needs to be controlled within a certain size. The texture features are described through the gray level gradient co-occurrence matrix, the gradient information of the image is fused after the position relation among the pixels is considered, the directionality of the texture can be better reflected, and the texture information of the gray level image can be more comprehensively represented. The gray-gradient co-occurrence matrix representation is of the form { Q (i, j); i 1, 2, … N,

j

1, 2, …, N, where each element represents the number of pixel points with a gray level of i and a gradient of j.

③ morphological feature extraction

The calibration of the defect actual area is completed by extracting corresponding fruit samples including the fruit samples with complete surfaces and defects on the surfaces, calculating the percentage of the defect area through the defect surface area and other morphological characteristic parameters according to the related standards of the fruit surface quality and the requirements of automatic fruit identification lines based on machine vision.

The actual area of the flaw can be converted by converting the formula S to be S pixel/767; and positioning and segmenting the defective part by utilizing image preprocessing to complete the extraction of morphological characteristics.

④ grade and variety judgment

By researching a classification method of a Gaussian mixture model, the fruit color and the area characteristics in the sample image are put into the Gaussian mixture model for training, and a corresponding fruit classifier is established. And then sequentially classifying the color and morphological region characteristics of the corresponding fruits by using a classifier. And finally, judging the final category of the fruits by combining the results of the two-time classification, and performing grade and variety identification according to the grade classification standard of the corresponding fruit.

Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing embodiments, or equivalents may be substituted for elements thereof. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. The utility model provides a fruit is multiaspect image acquisition device for classification which characterized in that: the fruit sample image digitization system comprises an image digitization processing system (1) for shooting images and realizing fruit sample image digitization, an upper PC (2) and an image processing and information analysis system (3) for analyzing surface quality and defect identification, wherein the image digitization processing system (1) comprises a camera (11) and a backlight plate (12) matched with the camera (11), the camera (11) is in signal connection with the upper PC (2) through a network cable, the image processing and information analysis system (3) is arranged in the upper PC (2), the image processing and information analysis system (3) comprises a feature extraction unit (31), and the feature extraction unit (31) comprises a color feature extraction unit (32) and a region feature extraction unit (33).

2. The multi-faceted image capturing apparatus for fruit sorting according to claim 1, wherein: the video camera (11) adopts a binocular CCD area array industrial camera.

3. The multi-faceted image capturing apparatus for fruit sorting according to claim 1, wherein: the color feature extraction unit (32) collects three color vectors of red, yellow and blue.

4. The multi-faceted image capturing apparatus for fruit sorting according to claim 1, wherein: the region feature extraction unit (33) acquires texture features and morphological features of an image.

5. The multi-faceted image capturing apparatus for fruit sorting according to claim 1, wherein: the upper PC (2) is internally provided with a median filter and a binomial filter for eliminating noise and Gaussian noise carried in the image.

6. The multi-faceted image capturing apparatus for fruit sorting according to claim 1, wherein: the picture digital processing system (1) is arranged on the fruit conveying device.