초록 ▼

The invention relates to a method, a sensor unit and a machine for detecting “sugar end” defects in potatoes. The method comprises irradiating potatoes with at least one light source, for a plurality of locus points on each potato, wherein locus points lie on the end regions of the potato and other

The invention relates to a method, a sensor unit and a machine for detecting “sugar end” defects in potatoes. The method comprises irradiating potatoes with at least one light source, for a plurality of locus points on each potato, wherein locus points lie on the end regions of the potato and other locus points lie in a central region of the potato. The light reflected from and/or transmitted through the respective locus points is selectively projected onto at least one photo sensor which generates light measurement signals for each locus point from the received light. At least one classification feature is determined from the light measurement signals. If at least one classification feature corresponds to a predefined “sugar end” criterion the respective potato is classified as having “sugar end” defects.

대표청구항 ▼

1. A method for detecting “sugar end” defects in potatoes, comprising: irradiating potatoes with at least one light source with wide-band light, for a plurality of locus points on each potato, wherein locus points lie on both end regions of the potato and other locus points lie in a central region o

1. A method for detecting “sugar end” defects in potatoes, comprising: irradiating potatoes with at least one light source with wide-band light, for a plurality of locus points on each potato, wherein locus points lie on both end regions of the potato and other locus points lie in a central region of the potato, selectively projecting the light reflected from or transmitted through the respective locus points onto at least one photo sensor, generating, with the at least one photo sensor, light measurement signals for each locus point from the received light which light measurement signals are generated as the spectra of the light reflected from or transmitted through the locus points, and intermediately storing such light measurement signals generated for each locus point, determining at least one classification feature from the light measurement signals in a wavelength range between 350 and 2500 nm by determining at least one classification feature from the spectra of the nth (n=1, 2, . . . ) derivative of the spectra, andclassifying potatoes as having “sugar end” defects if at least one classification feature corresponds to a predefined “sugar end” criterion,wherein determining at least one classification feature from the respective spectrum or the nth derivative of the respective spectrum of the locus points comprises calculating the difference curve for the respective locus point by calculating the differences between the spectral light measurement signals of the respective locus point and the spectral values of a reference spectrum for a number of wavelengths or by calculating the differences between the nth derivative of the spectral light measurement signals of the respective locus point and the nth derivative of the reference spectrum for a number of wavelengths, and that the classification feature is thus determined by the difference curve calculated in this way,wherein a classification feature derived from the difference curves is the curve of the first or second differential quotient of the respective difference curve and the classification criterion is selected from exceeding or falling below limit values of the differential quotient at defined wavelengths or wavelength bands, optionally with calculating average values of the differential quotient values within the wavelength bands, or the presence of difference quotient values within or out of a defined differential quotient value range at defined wavelengths or wavelength bands, optionally with calculating average values of the differential quotient values within the wavelength bands, or the similarity of at least one portion of the differential quotient curve with a predefined pattern. 2. A method according to claim 1, characterized in that the light reflected from or transmitted through the locus points is sequentially projected onto the at least one photo sensor. 3. A method according to claim 2, characterized in that the potatoes are moved along a conveying direction at a defined conveying velocity and that the light reflected from or transmitted through the locus points is projected successively onto the same photo sensor, spaced apart in time. 4. A method according to claim 3, characterized in that a plurality of photo sensor is arranged distributed transversely to the conveying direction, wherein—viewed in the conveying direction—locus points that are situated one after the other are each associated with the same photo sensor. 5. A method according to claim 1, characterized in that the light measurement signals contain light intensity values for a wavelength or wavelength band or contain several wavelengths or wavelength bands spaced apart, wherein the classification feature is these light intensity values and wherein the “sugar end” criterion is defined as the deviation from a threshold light intensity value. 6. A method according to claim 5, characterized in that the light intensity values are obtained at the particular wavelengths or wavelength bands by way of parallel or sequential irradiation of the potatoes with narrow-band light having the particular wavelengths or wavelength bands or by way of irradiation of the potatoes with wide-band light and filtering of the light reflected or transmitted at the respective locus points with band pass filters, the pass bands of which containing the particular wavelengths or wavelength bands or by way of provision of photo sensors, which are sensitive in the particular wavelengths or wavelength bands. 7. A method according to claim 1, characterized in that the respective light reflected from or transmitted through the locus points is received by splitting up the spectrum into a plurality of spectral components and that each spectral component is then projected onto one of the light-sensitive pixels associated with this spectral component of a photo sensor having a plurality of pixels, wherein the pixels of the photo sensor are optionally arranged in a two-dimensional way, wherein the first dimension represents the locus points and the second dimension represents the spectral components of the light. 8. A method according to claim 1, characterized in that a classification feature derived from the difference curves is the intensity curve of the respective difference curve over the wavelength and the classification criterion is selected from exceeding or falling below the intensity limit values at defined wavelengths or wavelength bands, optionally with calculating average values of the intensity values within the wavelength bands, or the presence of intensity values within or out of a defined intensity value range at defined wavelengths or wavelength bands, optionally with calculating average values of the intensity values within the wavelength bands, or the similarity of at least one portion of the intensity curve with a predefined pattern. 9. A method according to claim 1, characterized in that the reference spectrum is calculated as an averaged spectrum from the spectral image data of locus points of at least one potato, preferably from the spectral image data of locus points, which are associated with a centre region of one or several potatoes. 10. A method according to claim 9, characterized in that the reference spectrum is calculated or updated as a locally or timely averaged spectrum from the spectral image data of locus points of one or several potatoes to be examined in regard to “sugar end” defects, optionally while the potatoes are moved along the conveying direction. 11. A method according to claim 1, characterized in that determining at least one classification feature from the respective spectrum or nth derivative of the respective spectrum of the locus points comprises calculating the concentrations of ingredients, or evaluating the glassiness from the spectrum or from the nth derivative of the spectrum of the respective locus point, and that the classification features are selected from the determined concentration values of the ingredients or combinations thereof. 12. A method according to claim 11, characterized in that from the spectrum or the nth derivative of the spectrum of at least one locus point, wherein the locus point lies preferably in a central region of a potato, wherein optionally a plurality of locus points are used calculating the average value of their spectra or the nth derivative of the spectra, there are determined reference concentrations of ingredients or the glassiness, and that the deviation of the determined reference concentration values of the ingredients or combinations thereof or of the reference glassiness, from concentrations of ingredients or combinations thereof or from the glassiness at locus points, which lie on the end regions of potatoes, represent a “sugar end” criterion. 13. A method for detecting “sugar end” defects in potatoes, comprising: irradiating potatoes with at least one light source with wide-band light, for a plurality of locus points on each potato, wherein locus points lie on both end regions of the potato and other locus points lie in a central region of the potato, selectively projecting the light reflected from or transmitted through the respective locus points onto at least one photo sensor, generating, with the at least one photo sensor, light measurement signals for each locus point from the received light which light measurement signals are generated as the spectra of the light reflected from or transmitted through the locus points, and intermediately storing such light measurement signals generated for each locus point, determining at least one classification feature from the light measurement signals in a wavelength range between 350 and 2500 nm by determining at least one classification feature from the spectra or the nth derivative of the spectra, andclassifying potatoes as having “sugar end” defects if at least one classification feature corresponds to a predefined “sugar end” criterion,wherein determining at least one classification feature from the respective spectrum or the nth derivative of the respective spectrum of the locus points comprises calculating the difference curve for the respective locus point by calculating the differences between the spectral light measurement signals of the respective locus point and the spectral values of a reference spectrum for a number of wavelengths or by calculating the differences between the nth derivative of the spectral light measurement signals of the respective locus point and the nth derivative of the reference spectrum for a number of wavelengths, and that the classification feature is thus determined by the difference curve calculated in this way,wherein a classification feature derived from the difference curves is obtained from the at least portion-wise transformation of the difference curves into a visible wavelength range, thereby obtaining false-colour images, wherein preferably at least three portions of the difference curves are transformed into visible wavelength ranges, and the classification criterion is the locally defined appearance or absence of colours or colour ranges or colour transitions in the false-colour images.