WO2012074372A2

WO2012074372A2 - A system for fruit grading and quality determination

Info

Publication number: WO2012074372A2
Application number: PCT/MY2011/000237
Authority: WO
Inventors: Abdul Rashid Mohamed Shariff; Ahmad Rodzi Mahmud; Helmi Zulhaidi Mohd; Osama Mohamed Saeed; Mohd Din Amiruddin
Original assignee: Universiti Putra Malaysia (Upm)
Priority date: 2010-11-30
Filing date: 2011-11-25
Publication date: 2012-06-07
Also published as: BR112013013330A2; WO2012074372A3; MY164318A

Abstract

A system for fruit grading and quality determination characterized in that, the system (10) includes: a) housing (12) with an enclosure ( 12a) for scanning process; b) an illumination means (14) to provide shadow free illumination; c) a spectral camera ( 16) equipped with hyperspectral scanner together with a suitable charge-coupled device (CCD) array for capturing fruit sample's image; d) a conveying means ( 18) to provide scanning platform to the system; e) a processing unit (20) to process and analyze the fruit sample image; and e) a data acquisition interface (22) provided in between the spectral camera (16) and the processing unit (20); wherein the system utilizes hyperspectral imaging technology for agricultural product and quality inspections.

Description

A SYSTEM FOR FRUIT GRADING AND QUALITY DETERMINATION

FIELD OF INVENTION

The present invention generally relates to a system for fruit grading and quality determination, and more particularly to a system for fruit grading and quality determination of oil palm fruit by using hyperspectral imaging technology.

BACKGROUND OF THE INVENTION

A conventional practice of quality inspection, grading and safety control for agriculture crops or food harvesting is mainly performed by human inspections. The current practice in many oil palm mills today are using human inspections to grade oil palm bunches manually. ^However, these manual inspections of grading and quality assessment are labour intensive and time consuming. Moreover, the accuracy of grading results may be jeopardized by subjective human judgments. In addition, it is not easy for human to perform fast, easy and accurate grading and quality assessment of oil palm fruit bunches without using advance technology especially when performing large quantity inspection.

Innovative computer technologies have been employed in numerous applications where they have been incorporated with new machines for agricultural product grading and quality assessment. Some prior arts have brought out some inventions that disclosed quality inspection with various systems and methods.

For example, WO 2006/135267 A2 discloses a method and apparatus for monitoring fruit quality and ripeness using light-induced luminescence. Accordingly, this approach uses capability of volatile compounds produced by fruit or vegetables to luminescence in the surrounding areas of their surface when irradiated by a light of the suitable wavelength. WO 01 /69191 Al discloses an apparatus and method for measuring and correlating characteristics of fruit with visible/near infra-red spectrum. The invention discloses the utilization of the spectrum from 250 nm to 1 150 nm for measurement or prediction of one or more parameters. It also discloses the development of equipment and method of detecting emitted light from specimen exposed to the spectrum in at least one spectrum range, and in the preferred embodiment, in at least two spectrum ranges of 250 nm to 499nm and 500nm to 1 150nm. This disclosure relates generally to the use of the combined visible and near infrared spectrum in an equipment and method for measuring physical parameters.

US 2006/0203247 Al described how an apparatus and method are used for evaluating the interior quality of fruits and vegetables. Accordingly, the form of evaluation apparatus comprising a plurality of light source capable of irradiating measuring lights of given quality means for regulating the quality of measuring lights from the light sources in accordance with sizes of vegetables and fruits.

WO 99/60353 discloses a method of detecting colors by special categorization system to allow detection and sorting of object such as article whereas they move along a conveyor belt are well known. This system uses a spectral analysis of the article obtained by image camera. The system was employed to sort and refuse fruit and vegetables by their color to indicative the ripeness and to sort bottles for recycling by glass or plastic color.

Furthermore, US 5, 589, 209 discloses a method for non-destructive determination of quality parameters in fresh produce in which ultrasonic excitation is transmitted along a sound path through the tissue of fresh products such as fruits or vegetables. Sound waves are then detected and analyzed to obtain the sound speed and reduction which shows the quality parameters of the fresh product.

However, none of the above disclosures accommodate the requirements of oil palm fruit ripeness grading and quality assessment based on the special parameters and properties of oil palm fruits. A conventional imaging system provides limited waveband information, which restrains its capacity for detecting and providing information of an entire sample for grading and quality assessment. Since the oil palm fruits have specific features and the ripeness classes (under ripe, ripe and over rip) which are very close to each other, therefore specification of illumination and hyperspectral imaging system need to be highly considered.

A system of fruit grading and quality determination of the present invention focuses on specific techniques to work with the parameters and properties of oil palm fresh fruit bunch (FFB) by using hyperspectral imaging technology so that to obtain a fast, easy and accurate classification of oil palm fruit. The system and it combination of elements and/or arrangement of parts will be exemplified in the detailed description.

SUMMARY OF THE INVENTION

The present invention relates to a system for fruit grading and quality determination of oil palm fruit by using hyperspectral imaging technology. Accordingly, the system includes: a) housing with an enclosure for scanning process; b) an illumination means to provide shadow free illumination; c) a spectral camera equipped with hyperspectral scanner together with a suitable charge-coupled device (CCD) array for capturing fruit sample's image; d) a conveying means to provide scanning platform to the system; e) a processing unit to process and analyze the fruit sample image; and e) a data acquisition interface provided in between the spectral camera and the processing unit; wherein the system utilizes hyperspectral imaging technology for agricultural product and quality inspections.

In accordance with preferred embodiments of the present invention, the housing of the system is provided with sufficient dark enclosed space and environment for scanning process. The enclosure of the housing is formed by darken finishing material. It will be appreciated that the housing further provided with flexible doors with darken finishing material for controlling the system scanning environment.

The illumination means of the system can be of halogen lamps or any illumination source such as applied security design (ASD) lamps. Accordingly, the illumination means are disposed within the housing and are evenly distributed across sample area to produce illumination field in the system.

It will be appreciated that the spectral camera is provided at top portion of the housing and enclosed by a compartment. The spectral camera captures image information of fruit sample that is illuminated by predetermined visible band of light beam. It will appreciated that the spectral camera is further provided with objective lens in a path of light beam selectively controllable to pass by a predetermined visible wavelength band of the entire light beam. The spectral camera collects information as a set of images which represents electromagnetic spectrum or spectral band. The images are combined and form a three dimensional (3D) hyperspectral cube for processing and analysis. In the preferred embodiments, the conveying means is preferably furnished with non- reflective finishing for carrying the fruit samples into the illumination field of the system. The conveying means further includes a programmable motor to move the conveying means under a certain size and weight of fruit samples.

It will be appreciate that the processing unit (20) of the system is preferably a laptop computer. The processing unit process data information by hyperspactral data processing software such as ENVI^® software to perform fruit sample classification and to obtain denoised image of the fruit sample. The processing unit also used data processing software such as MATLAB^® to perform the analysis of the fruit sample classification and to obtain the resultant quality data.

The data acquisition interface in accordance with preferred embodiments of the system is a frame link and cable for data transferring, such as cable camera and frame grabber.

It is to be noted that the system is subjected to an Artificial Neural Network (ANN) technique for purposes of oil palm FFB ripeness classification. The technique of ripeness classification of oil palm FFB image is done by analysis of the fruit sample of three different ripeness categories, i.e. under ripe, ripe and over ripe. It will be appreciated that the system can be made portable or adopted in industrial chain framework by taking into consideration of fruit size, weight, and shape that makes the system multipurpose for use in similar application with different agriculture fruits. BRIEF DESCRIPTION OF THE DRAWINGS

FIGS, la and lb show a physical setup of a system for fruit grading and quality determination in accordance with preferred embodiment of present invention;

FIG. 2 shows illumination setup of the system in accordance with preferred embodiment of the present invention;

FIG. 3 is a work process flowchart of the system for fruit grading and quality determination of oil palm fruit;

FIG. 4 shows a plot of distinctive reflectance value versus wavelength of oil palm fruit in three different ripeness categories, i.e. under ripe, ripe and over ripe.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

A detailed description of preferred embodiments of the invention is disclosed herein. It should be understood, however, the disclosed preferred embodiments are merely exemplary of the invention, which may be embodied in various forms. Therefore, the details disclosed herein are not to be interpreted as limiting, but merely as the basis for the claims and for teaching one skilled in the art of the invention.

The system of fruit grading and quality determination of the present invention is focused on specific techniques to work with the parameters and properties of oil palm fresh fruit bunch (FFB) by using hyperspeciral imaging technology. It will be appreciated that different grading systems and quality assessments can also be performed with different techniques for different types of agriculture products. The system uses hyperspeciral imaging technology that provides wavelength determination of quality parameter and maturity grading of oil palm FFB. It also provides improved imaging system by reducing image noise.

The system for fruit grading and quality determination of the present invention will now be described in detail in accordance to the accompanying drawings, FIGS. 1 to 4, both individually and in any combination thereof.

With referring to FIGS, la and lb, the system for fruit grading and quality determination (10) of the present invention generally includes a housing (12), illumination means (14), spectral camera (16), conveying means (18), processing unit (20) and data acquisition interface (22).

Accordingly, the housing (12) of the present invention is provided with an enclosure (12a) for scanning process. The housing (12) is important for indoor grading system to provide sufficient dark enclosed space and environment for scanning process. It will be appreciated that the enclosure (12a) is formed by darken finishing material. Preferably, said darken finishing material is of non-reflective black colour finishing. A flexible door (12b) with darken finishing material is also provided to control the system scanning environment. Optionally, the flexible door (12b) can be of, but not limited to a flexible rubber strap material.

In the preferred embodiments, the illumination means (14) are provided in the housing (12). The illumination means (14) can be of, but not limited to halogen lamps or any illumination source such as applied security design (ASD) lamps to provide shadow free illumination. Preferably, the halogen lamps are used in the system. It will be appreciated that the illumination means (14) are disposed within the housing (12) and are evenly distributed across sample area to produce illumination field in the system as shown FIG. 2.

In the preferred embodiments, the spectral camera (16) is provided at top portion of the housing (12) and enclosed by a compartment (13). Particularly, the spectral camera (36) is positioned up-right on top of the housing (12) to capture image information of the fruit sample that illuminated by predetermined visible band of light beam. It is to be noted that the spectral camera (16) is equipped with hyperspectral scanner together with a suitable charge-coupled device (CCD) array for capturing fruit sample's image. The hyperspectral scanner is used for sensing variety of light spectral. Objective lens (not shown) may be further provided in the path of light beam selectively controllable to pass by a predetermined visible wavelength band of the entire light beam. Accordingly, the objective lens is act as illumination filter to the system. By way of example but not limitation, an OLE23 objective lens is preferably used.

The spectral camera (16) with hyperspectral scanner enables the scanning and captures the fruit sample data entirely when the sample is conveyed across the sample area (17) of the housing (12) by conveying means (18). The spectral camera (16) collects information as a set of images which represents electromagnetic spectrum which is also known as spectral band. The images are then combined and form a three dimensional (3D) hyperspectral cube for processing and analysis. It will be appreciated that scanning time for one oil palm fruit bunch depends on the integration time used for the spectral camera (16) and the size of the fruit. The conveying means (18) is used for conveying the fruit sample to sample area (17) of the housing (12). It will be appreciated that a programmable motor (19) is provided to generate power to control and to move the conveying means (18) under a certain size and weight of fruit samples. The conveying means (18) is preferably furnished with non-reflective finishing for carrying the fruit samples into the illumination field of the system. The conveying means (18) also provide a scanning platform to the system. Hyperspectral image of the fruit sample will then be captured by the spectral camera (16). The processing unit (20), preferably a laptop computer is provided to process and analyze the fruit sample image. It will be appreciated that after capturing and scanning processes of one entire fruit sample, for instant oil palm FFB (21 ), spatial- by-spectral matrices of hypespectral imaging will be combined to construct a three dimensional (3D) spatial and spectral data space.

The data acquisition interface (22) is provided in between the spectral camera (16) and the processing unit (20) by a frame link and cable for data transferring. Preferably, cable camera and frame grabbers are used to transfer data. The data information will be processed by the processing unit (20). Preferably, hyperspactral data processing software, for example, ENVI^® software is used to perform classification such as subseting, image resizing, and minimum noise fraction (MNF) to remove the noise from the image. Denoised image will then be used with data processing software such as MATLAB^® to perform the analysis of the fruit sample classification and to obtain the resultant quality data.

It will be appreciated that data collected by the system is subjected to the Artificial Neural Network (ANN) technique for purposes of oil palm FFB ripeness classification. Accordingly, technique of ripeness classification of oil palm FFB image is done by analysis of the fruit samples of three different ripeness categories, i.e. under ripe, ripe and over ripe. For example, the resultant data can be plotted as shown in FIG. 4, wherein distinctive reflectance value versus wavelength of oil palm fruits of are plotted three different ripeness categories, i.e. under ripe, ripe and over ripe. It will be appreciated that the system can be setup for use with other different fruits sample of similar application.

It will also be appreciated that the system of the present invention provides nondestructive measurement techniques of assessment which has tremendous applications in the agriculture and food industry, including the inspection and grading of vegetables and fruits. The system design can be made portable or adopting in industrial chain framework by taking into consideration of fruit size, weight, and shape that make the system multipurpose for use in similar application with different agriculture fruits.

It is to be noted that setting of the system is very important to obtain good accuracy. It is within the system setup of the spectral camera. This can be done by connecting USB connector and PCMCIA card (not shown) to the assigned laptop computer and the hyperspectral scanner. White reference tile is then installed and a calibration sheet on x-stage scanning platform is setup as shown in FIG. 2. The illumination means is to be turned on to ensure even distribution.

It will be appreciated that spectral imaging properties of the spectral camera will be automatically launched to prompt on selection of desired band file by spectralDAQ software installed in the laptop computer for spectral camera controls frame adjustment, exposure time and for monitoring profile plot peak within saturation range. By way of example and not by way of limitation, the height of the spectral camera is adjusted from x-stage scanning platform, accordingly at least 1.1m from object height for the objective lens and the focus of spectral camera is adjusted by using a calibration sheet to ensure the sharpness of the image.

It will also be appreciated that the hyperspectral scanner can be set within setting system of the spectralDAQ software. Particularly on setting scan mirror control for determining the start and end points along the sample area. Scan rate can be adjusted by visually comparing the ratio of the actual length : width with the viewed image. With all of these steps the spectral camera setup is ready for measurement of the oil palm fruit bunch. Before measuring any fruit sample, one white and two dark references are always measured. The dark reference used to remove the effect of dark current of CCD detectors, which are thermally sensitive. The oil palm fruit bunch is place on the white reference on the scanning platform to define the start and end positions before running the scanner.

After the scan of one entire oil palm fruit bunch is completed, the spatial-by-spectral matrices are combined to construct a three dimensional (3D) spatial and spectral data space. As mentioned earlier, the scanning time for one oil palm bunch depends on the integration time used for the camera and the size of the fruits. After obtaining the data, the hyperspectral data processing software such as ENVI^® software is used to do classification such as subseting, image resizing, minimum noise fraction MNF to remove the noise from the image. The denoised image is used with data processing software such as MATLAB^® to do the analysis of the FFB classification. Work process of the system for fruit grading and quality determination of oil palm fruit is summarized in the flowchart as shown in FIG. 3.

While embodiments of the invention have been illustrated and described, it is not intended that these embodiments illustrate and describe all possible forms of the invention. Rather, the words used in the specification are words of description rather than limitation and various changes may be made without departing from the scope of the invention.

Claims

A system for fruit grading and quality determination characterized in that, the system (10) includes:

a) a housing (12) with an enclosure (12a) for scanning process;

b) an illumination means (14) to provide shadow free illumination;

c) a spectral camera (16) equipped with hyperspectral scanner together with a suitable charge-coupled device (CCD) array for capturing fruit sample's image;

d) a conveying means (18) to provide scanning platform to the system; e) a processing unit (20) to process and analyze the fruit sample image; f) a data acquisition interface (22) provided in between the spectral camera (16) and the processing unit (20);

wherein the system utilizes hyperspectral imaging technology for agricultural product and quality inspections.

A system for fruit grading and quality determination according to Claim 1, wherein the housing (12) is provided with sufficient dark enclosed space and environment for scanning process.

A system for fruit grading and quality determination according to Claim 1, wherein the enclosure (12a) of the housing (12) is formed by darken finishing material.

A system for fruit grading and quality determination according to Claim 1, wherein the housing (12) further provided with flexible doors (12b) with darken finishing materia] for controlling the system scanning environment.

5. A system for fruit grading and quality determination according to Claim 1, wherein the illumination means (14) can be of halogen lamps or any illumination source such as applied security design (ASD) lamps.

A system for fruit grading and quality determination according to Claim 1 , wherein the illumination means (14) are disposed within the housing. (12) and are evenly distributed across sample area to produce illumination field in the system.

A system for fruit grading and quality determination according to Claim 1 , wherein the spectral camera (16) is provided at top portion of the housing (12) and enclosed by a compartment (13).

A system for fruit grading and quality determination according to Claim 1, wherein the spectral camera (16) captures image information of fruit sample that is illuminated by predetermined visible band of light beam.

A system for fruit grading and quality determination according to Claim 1, wherein the spectral camera (16) is further provided with objective lens in a path of light beam selectively controllable to pass by a predetermined visible wavelength band of the entire light beam.

A system for fruit grading and quality determination according to Claim 1 , wherein the spectral camera (16) collects information as a set of images which represents electromagnetic spectrum or spectral band.

A system for fruit grading and quality determination according to Claim 10, wherein the images are combined and form a three dimensional (3D) hyperspectral cube for processing and analysis.

A system for fruit grading and quality determination according to Claim 1 , wherein the conveying means (18) is preferably furnished with non-reflective finishing for carrying the fruit samples into the illumination field of the system.

13. A system for fruit grading and quality determination according to Claim 1 , wherein the conveying. means (18) further includes a programmable, motor (19) to move the conveying means (18) under a certain size and weight of fruit samples.

14. A system for fruit grading and quality determination according to Claim 3 , wherein the processing unit (20) is preferably a laptop computer.

15. A system for fruit grading and quality determination according to Claim 1, wherein the processing unit (20) process data information by hyperspactral data processing software such as ENVI^® software to perform fruit sample classification and to obtain denoised image of the fruit sample.

16. A system for fruit grading and quality determination according to Claim 1. wherein the processing unit (20) used data processing software such as MATLAB^® to perform the analysis of the fruit sample classification and to obtain the resultant quality data.

17. A system for fruit grading and quality determination according to Claim 1, wherein the data acquisition interface (22) is a frame link and cable for data transferring, such as cable camera and frame grabber.

18. A system for fruit grading and quality determination according to Claim 1, wherein the system (10) is subjected to an Artificial Neural Network (ANN) technique for purposes of oil palm FFB ripeness classification.

19. A system for fruit grading and quality determination according to Claim 18, wherein the technique of ripeness classification of oil palm FFB image is done by analysis of the fruit sample of three different ripeness categories, i.e. under ripe, ripe and over ripe. A system for fruit grading and quality determination according to Claim 1 , wherein the system ( 10) can be made portable or adopted in industrial chain framework by taking into consideration of fruit size, weight, and shape that makes the system multipurpose for use in similar application with different agriculture fruits.