阅读说明：本技术 一种基于非模型光学校正高光谱的食品检测方法 (A kind of food detection method based on non-model optical correction EO-1 hyperion ) 是由 孙大文 马骥 蒲洪彬 韦庆益 于 2019-05-07 设计创作，主要内容包括：本发明公开了一种基于非模型光学校正高光谱的食品检测方法,包括以下步骤：S1准备参考背景食品和参考富集食品；S2采集参考背景食品和参考富集食品的散射高光谱图像作为参考标准；S3得到参考背景食品和参考富集食品的参考曲线特征值的三维数据；S4得到待测食品的散射高光谱图像为三维数据；S5得到待测物含量的校正结果。本发明的食品检测方法用散射距离曲线每一点的曲线特征值作为特征,以标准品光学特征作为参考计算,无需进行复杂的数据计算或建立复杂的数学预测模型,具有较高的预测精度和很好的稳定性。(The invention discloses a kind of food detection methods based on non-model optical correction EO-1 hyperion, comprising the following steps: S1 prepares with reference to background food and refers to enriched food；S2 acquisition is with reference to the scattering high spectrum image of background food and reference enriched food as reference standard；S3 is obtained with reference to background food and the three-dimensional data of the reference curve characteristic value with reference to enriched food；The scattering high spectrum image that S4 obtains food to be measured is three-dimensional data；S5 obtains the correction result of determinand content.Food detection method of the invention uses the eigenvalue of curve of scattering distance Curve every bit as feature, using standard items optical signature as with reference to calculating, complicated mathematical forecasting model, precision of prediction with higher and good stability are calculated or established without carrying out complicated data.)

1. a kind of food detection method based on non-model optical correction EO-1 hyperion, it is characterised in that the following steps are included:

S1 prepares with reference to background food and determines with reference to background food determinand content V_b；Prepare with reference to enriched food and determines ginseng Examine enriched food determinand content V_r；

S2 moveable stage is static, and the linear light source of power adjustable is gradually decreased from the initial power higher than operating power makes light intensity Linear reduction when power drops to 0, obtains the higher reference with reference to food of reflectivity and scatters high spectrum image, determine therein Operating power is power when detecting tested food；

Linear light source power is adjusted, keeps reflectivity lower with reference to reflectivity under food initial scatter spectral maximum and initial power The higher scattering spectrum maximum value with reference to food is equal, and objective table is static, and power, which gradually decreases, makes light intensity linear reduction, power When dropping to 0, obtains the lower reference with reference to food of reflectivity and scatter high spectrum image；

The reference food is with reference to background food or to refer to enriched food；

S3 provides in scattering high spectrum image to calculate the scattering for referring to background food by the boundary of close to sources to scatter apart from zero point The eigenvalue of curve C of distance Curve every bit_bWith the eigenvalue of curve C of the scattering distance Curve every bit of reference enriched food_r；

Respectively obtain the corresponding light intensity L of three-dimensional data with reference to background food spectral dispersion value_bWith reference enriched food spectral dispersion The corresponding light intensity L of the three-dimensional data of value_r；

S4 objective table is moved with fixed speed, and linear light source is adjusted to operating power and power is constant always, acquires article to be measured High spectrum image is scattered, the eigenvalue of curve C of the scattering distance Curve every bit of article to be measured is calculated_s；

S5 is to C_s、C_b、C_r、V_b、V_rFunctional relation F (C) is established, and predicts the determinand content of each pixel；With containing L_bAnd L_r Functional relation F (L) correction F (C) function as a result, obtaining the correction result F (LC) of determinand content.

2. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: step In rapid S1, known to without determinand or determinand content and food that content is extremely low is used as and refers to background food；Containing only to be measured Object or determinand content is known and the very high food of content is used as and refers to enriched food.

3. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: step In rapid S2, the linear light spot of the linear light source of the power adjustable is vertical with the scan line of line scan camera.

4. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: step In rapid S2, the linear light source of the power adjustable generates spot width and adjusts in 1-5mm.

5. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: step In rapid S2, the linear light source of the power adjustable is adjusted at a distance from the line scan camera visual field in 1-5mm.

6. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: step In rapid S4, objective table moving direction is vertical with line scanning high spectrum image acquisition system gathering line.

7. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: anti- Penetrate the higher reflectivity for referring to background food and reference enriched food by comparing with reference to food under the conditions of same light is shone of rate It obtains.

8. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: right It takes off in pig as food to be measured, to the first derivative absolute value C for the scattering distance Curve every bit that pig takes off_s, pork longissimus dorsi muscle Scattering distance Curve every bit first derivative absolute value C_b, fat scattering distance Curve every bit first derivative it is absolute Value C_r, pork longissimus dorsi muscle fat content V_b, pork fat fat content V_rEstablish functional relation F (C):To the corresponding light intensity L of three-dimensional data of pork longissimus dorsi muscle spectral dispersion value_bWith pork fat light Compose the corresponding light intensity L of three-dimensional data of scattering value_rEstablish functional relationF (C) value is multiplied with F (L) value, Obtain correction function F (LC).

9. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: right In banana as food to be measured, new fresh bananas, which is used as, refers to background food, and artificial damage banana, which is used as, refers to background food；To new The second dervative absolute value C of the scattering distance Curve every bit of fresh bananas_b, artificial damage banana scattering distance Curve every bit Second dervative absolute value C_r, with reference to enriched food browning degree V_r, with reference to background food browning degree V_b, banana to be measured dissipate Penetrate the second dervative absolute value C of distance Curve every bit_sEstablish functional relation

Light intensity L corresponding to the three-dimensional data of new fresh bananas spectral dispersion value_bWith three dimensions of artificial damage banana spectral dispersion value According to corresponding light intensity L_rEstablish functional relation

F (C) value is multiplied with F (L) value, obtains correction function F (LC).

10. the food detection method according to claim 1 based on non-model optical correction EO-1 hyperion, it is characterised in that: It further include indicating the correction result of determinand content with visualization distribution map after step S5.

Technical field

The present invention relates to food field of non destructive testing, in particular to the food inspection based on non-model optical correction EO-1 hyperion Device and method.

Background technique

Food Quality and Safety problem is all the focus of public concern all the time.In recent years, non-destructive testing technology is wide General to apply in food inspection, wherein optical non-destructive detection technology is using more universal technology.In optical detection, it is seen that Near infrared band can provide the appearance characteristics and chemical information of food, and the optical detecting method based on this wave band section is main There are the methods of computer vision, near infrared spectrum, high light spectrum image-forming, these methods cut both ways, and have his own strong points.

Computer vision technique is suitble to detect the apparent food quality of external appearance characteristic, freshness, fruits and vegetables including meat Visual grading etc., but quality stronger for chemical information correlation can not then detect.Near-infrared spectrum technique can detecte The chemical information of food, the albumen fat content of moisture content, meat including food, sugar acid content of fruits and vegetables etc., but this skill The drawbacks of art is cannot to detect to the spatial distribution of food quality.Traditional hyperspectral technique can combine above two method Advantage, not only can detecte the chemical information of food, but also its spatial distribution and external appearance characteristic attribute can be provided, but its model It establishes a large amount of, multiplicity, the reliable sample of initial stage needs to be supported, early development cost is huge.Three of the above method exists When regression analysis or pattern-recognition, complicated algorithm is required as support.

Itself for hyperspectral technique, traditional hyperspectral technique has only used the reflection, absorption or scattering value of spectrum, but Have ignored the optical characteristics that light is propagated in food.Although tradition scattering hyperspectral technique analyzes optical characteristics, but it is empty Between information expression more lack, the information of " line " in scattering distance, both line scan camera linear array detector can only be detected The range of acquisition, while the demand modeled early period to a large amount of samples can not be got rid of.

Summary of the invention

In order to overcome the disadvantages mentioned above and deficiency of the prior art, the purpose of the present invention is to provide be based on non-model optical school The food inspection device and method of positive EO-1 hyperion, detection accuracy are high.

The purpose of the present invention is achieved through the following technical solutions:

A kind of food detection method based on non-model optical correction EO-1 hyperion, comprising the following steps:

S1 prepares with reference to background food and determines with reference to background food determinand content V_b；Prepare with reference to enriched food and true Surely enriched food determinand content V is referred to_r；

S2 moveable stage is static, and the linear light source of power adjustable is gradually decreased from the initial power higher than operating power to be made Light intensity linear reduction when power drops to 0, obtains the higher reference with reference to food of reflectivity and scatters high spectrum image, determine it In operating power be power when detecting tested food；

Linear light source power is adjusted, keeps reflectivity lower with reference to anti-under food initial scatter spectral maximum and initial power Penetrate that the higher scattering spectrum maximum value with reference to food of rate is equal, and objective table is static, and power, which gradually decreases, makes light intensity linear reduction, When power drops to 0, obtains the lower reference with reference to food of reflectivity and scatter high spectrum image；

The reference food is with reference to background food or to refer to enriched food；

S3 provides in scattering high spectrum image to calculate to scatter apart from zero point with reference to background food by the boundary of close to sources Scatter the eigenvalue of curve C of distance Curve every bit_bWith the curvilinear characteristic of the scattering distance Curve every bit of reference enriched food Value C_r；

Respectively obtain the corresponding light intensity L of three-dimensional data with reference to background food spectral dispersion value_bWith reference enriched food light Compose the corresponding light intensity L of three-dimensional data of scattering value_r；

S4 objective table is moved with fixed speed, and linear light source is adjusted to operating power and power is constant always, acquires determinand The scattering high spectrum image of product calculates the eigenvalue of curve C of the scattering distance Curve every bit of article to be measured_s；

S5 is to C_s、C_b、C_r、V_b、V_rFunctional relation F (C) is established, and predicts the determinand content of each pixel；With containing L_bAnd L_rFunctional relation F (L) correction F (C) function as a result, obtaining the correction result F (LC) of determinand content.

To further realize the object of the invention, it is preferable that in step S1, known to without determinand or determinand content and The extremely low food of content, which is used as, refers to background food；Containing only determinand or determinand content known to and the very high food of content make For with reference to enriched food.

Preferably, in step S2, the linear light spot of the linear light source of the power adjustable and the scan line of line scan camera are hung down Directly.

Preferably, in step S2, the linear light source of the power adjustable generates spot width and adjusts in 1-5mm.

Preferably, in step S2, the linear light source of the power adjustable is adjusted at a distance from the line scan camera visual field in 1- 5mm。

Preferably, in step S4, objective table moving direction is vertical with line scanning high spectrum image acquisition system gathering line.

Preferably, reflectivity is higher refers to background food and with reference to enriched food in same light with reference to food by comparing Reflectivity according under the conditions of obtains.

It takes off preferably for pig as food to be measured, the first derivative for the scattering distance Curve every bit taken off to pig is absolute Value C_s, pork longissimus dorsi muscle scattering distance Curve every bit first derivative absolute value C_b, fat scattering distance Curve it is each The first derivative absolute value C of point_r, pork longissimus dorsi muscle fat content V_b, pork fat fat content V_rEstablish functional relation F (C):To the corresponding light intensity L of three-dimensional data of pork longissimus dorsi muscle spectral dispersion value_bWith pork rouge The corresponding light intensity L of the three-dimensional data of fat spectral dispersion value_rEstablish functional relationBy F (C) value and F (L) value It is multiplied, obtains correction function F (LC).

Preferably for banana as food to be measured, new fresh bananas, which is used as, refers to background food, artificial damage banana conduct With reference to background food；To the second dervative absolute value C of the scattering distance Curve every bit of new fresh bananas_b, artificial damage banana Scatter the second dervative absolute value C of distance Curve every bit_r, with reference to enriched food browning degree V_r, with reference to background food brown stain journey Spend V_b, banana to be measured scattering distance Curve every bit second dervative absolute value C_sEstablish functional relation

Light intensity L corresponding to the three-dimensional data of new fresh bananas spectral dispersion value_bWith artificial damage banana spectral dispersion value The corresponding light intensity L of three-dimensional data_rEstablish functional relation

F (C) value is multiplied with F (L) value, obtains correction function F (LC).

It preferably, further include indicating the correction result of determinand content with visualization distribution map after step S5.

Compared with prior art, the present invention has the following advantages and beneficial effects:

(1) present invention scatters high spectrum image with linear light source type collection, has obtained three-dimensional scattering high spectrum image, than passing The information that the two-dimensional scattering high spectrum image of system includes is richer.

(2) present invention uses the eigenvalue of curve of scattering distance Curve every bit as feature, with standard items optical signature work To obtain a large amount of modeling sample without carrying out a large amount of experiment with reference to calculating.

(3) present invention is analyzed using simple specific function, calculates or establish complicated number without carrying out complicated data Learn prediction model, precision of prediction with higher and good stability.

Detailed description of the invention

Fig. 1 is the schematic diagram that the food inspection device of EO-1 hyperion is corrected the present invention is based on non-model optical.

Fig. 2 is the pork fat content and marbling detection that EO-1 hyperion is corrected based on non-model optical of embodiment 1 The flow chart of method.

Fig. 3 is the banana browning degree and distribution detection method that EO-1 hyperion is corrected based on non-model optical of embodiment 2 Flow chart.

Specific embodiment

For a better understanding of the invention, the present invention is further illustrated with reference to the accompanying drawings and embodiments, but this hair Bright embodiment is without being limited thereto.

Device shown in FIG. 1 is for realizing the food detection method based on non-model optical correction EO-1 hyperion.Such as Fig. 1 institute Show, high spectrum image acquisition system is scanned based on the food inspection device of non-model optical correction EO-1 hyperion, including darkroom A, line B, power adjustable linear light source C, moveable stage D and computer E；Line scans high spectrum image acquisition system B, power adjustable line Light source C and moveable stage D is arranged in the A of darkroom, the online scanning high spectrum image acquisition system of moveable stage D setting The lower end B, power adjustable linear light source C setting can be irradiated in the side moveable stage D, the light source of power adjustable linear light source C transmitting On the article to be measured of moveable stage D.