JP3311880B2 - Automatic detection device for fruits and vegetables injury - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for judging the appearance quality of fruits and vegetables, which has been conventionally performed by a skilled inspector.
Typically, the judgment of the presence or absence and size of injuries is automatically,
The present invention relates to a device for automatically detecting injury of fruits and vegetables which can be performed mechanically.

[0002]

BACKGROUND OF THE INVENTION Appearance quality, which is one of the judging factors when sorting or evaluating fruits and vegetables, is based on surface conditions such as color, presence or absence of injury, etc., in addition to classification by shape and size. Grading is also important, and it has a significant weight that cannot be ignored as a factor for sorting fruits and vegetables in the fruit selection area of the production area.

[0003] Among such factors such as classification, relatively simple components such as the shape and size of fruits and vegetables can be evaluated relatively easily based on image-capturing information from a monochrome camera or the like. It is actually widely used for automatic sorting as well as sorting elements such as weighed weight.

[0004] However, grading based on factors such as the presence or absence of injuries and colors of fruits and vegetables is difficult despite the fact that it is difficult to secure skilled workers, and there is a demand for cost reduction by automation and mechanization of work. In fact, the performance of the conventionally proposed automation methods and apparatuses is remarkably inferior to the level of judgment by a skilled worker, and there are many problems, and no appropriate solution has been provided yet. .

[0005] Conventional proposals for a method of automatically judging the presence or absence of an injury include, for example, a method of detecting an injury of "heavy fruits and vegetables" or a method of detecting a citrus white wound (Japanese Unexamined Patent Publication No. No. 218944) is known.

[0006] The former method of detecting a flaw in a "fat-and-vegetable fruit" is based on calculating saturation from R, G, and B signals obtained by capturing a planar image of a fruit and vegetable with a color camera, and fixing the flat image. After removing the inside of the diameter circle as the center part with a letter, and removing the peripheral part to prevent erroneous detection due to blurring around the image or uneven illumination, the saturation of the remaining part is a fixed reference value In comparison with, a point lower than the reference value is determined as a flaw.

In the latter method of detecting citrus white flaws, a citrus skin surface is photographed with a camera having a sensitivity in a wavelength band excluding the orange color, which is a citrus skin color, and wavelengths in the vicinity of the orange color. Is subjected to a binarization process at a first binarization level higher than the luminance level of the citrus skin color and lower than the luminance level of the white lesion, and based on the binarization signal (white flaw + spot portion) And a binarization process is performed at a second binarization level higher than the luminance level of the white lesion of the video signal and lower than the luminance level of the print area, and the binarization is performed. Based on the signal (segment) only the second
Is obtained by subtracting the second number of pixels from the first number of pixels to determine the number of pixels only white lesions,
Detection is performed using light in a specific wavelength band and utilizing the difference in brightness between the white flaw and the print area.

[0008]

However, in the above-mentioned method for detecting the injury of "heavy fruits and vegetables", the head portion is uniformly determined as the center of a circle having a constant diameter. Given that the position of the letter of the fruit is not always constant, it is necessary to determine the diameter of the circle to be excluded relatively large, and to remove the peripheral part due to uneven lighting etc. The area is extremely narrow and limited, so that it is only possible to detect an injury that is extremely inadequate as compared with the level of judgment by a skilled worker, and cannot satisfy practical use at all.

In the latter method for detecting white flaws on citrus fruits, the white flaws are detected by light / dark processing using luminance using illumination light in a specific wavelength range. Not detectable. Therefore, in order to automate the judgment of the class including those other than the white scratches, it is necessary to develop another injury detecting means.

[0010] By the way, the detection of injury in fruits and vegetables is as follows.
As described above, it is not only necessary to combine the optical technology of illumination and imaging with the signal processing technology obtained thereby, but also the fruits and vegetables to be detected have different colors or different types of injuries. It is difficult to say that the type, color, etc., are constant.
It is difficult to consider a versatile injury detection method that can be applied to various fruits and vegetables because there is a great variety of fruits and fruits and vegetables of the same variety may have different colors depending on the season and the like. Even if a dedicated machine for a single product is provided, it is not easy to solve the problem.

For example, considering only the point of obtaining an imaging signal for signal processing, it is not enough to image only the upper surface of the fruits and vegetables to detect a flaw in the widest possible range of the fruits and vegetables surface. Although it is conceivable to photograph a wide area using a camera, the conventional method of detecting injuries by comparing with a certain reference value may cause erroneous detection of injuries if the sensitivity adjustment between multiple cameras used is not sufficient. Inevitable. Further, in the above-described conventional method of detecting a white flaw by light / dark processing based on luminance, the problem of erroneous detection caused by fluctuation of the light amount of the light source over time and uneven illumination cannot be ignored.

As described above, the conventional method and apparatus in which injury judgment is automated and mechanized are practically impossible to be practically used on an industrial level, or are possible in an extremely limited narrow range and limited items. In practice, it was only possible to judge the appearance quality such as the presence or absence of injury by a skilled worker at a level that was not very accurate. However, the present invention automatically and mechanically detects such injury of fruits and vegetables. In addition, the present invention has been made in order to provide a device that can perform the determination accurately and stably.

[0013] That is, an object of the present invention is to provide an automatic injury detection apparatus for fruits and vegetables that can detect an injury stably and with high accuracy over a wide range of the epidermis of fruits and vegetables.

Another object of the present invention is to minimize the erroneous detection of injuries due to the influence of temporal changes in the amount of light of the light source and uneven illumination of the illumination system, which is close to the judgment level of a skilled worker. It is an object of the present invention to provide a fruit and vegetable injury automatic detection device capable of performing high-level automation and mechanized judgment.

Still another object of the present invention is to provide a versatility which can be changed so as to be able to perform a specified injury detection on fruits and vegetables of different varieties, and which can be widely used for injury detection on a large variety of fruits and vegetables. To provide an automatic apparatus for automatically detecting injuries of fruits and vegetables with high cost.

[0016]

One of the features of the automatic fruit and vegetable injury detecting apparatus according to the present invention which achieves the above object is that a light source for illuminating a plurality of fruits and vegetables conveyed with a distance. Means, an image of the illuminated fruits and vegetables, and light receiving means for extracting red, green, and blue signals for each pixel, and calculating the saturation (C) of each pixel based on the red, green, and blue signals. Calculating means for calculating a partial average value of the surrounding saturation for each pixel by a moving average method, and comparing the saturation of each pixel with the corresponding partial average value to determine the presence or absence of an injury-corresponding part. And a determination means.

In determining the presence or absence of an injury, it is preferable to provide an image of a fruit or vegetable table as a target as wide as possible. For this purpose, a method of imaging three surfaces with a single camera ( Japanese Patent Publication No. 5-60544), a method of imaging five surfaces with a single camera (Japanese Patent Laid-Open No. 63-611)
No. 05) can be used.
Even when these methods are employed, it is possible to accurately determine the presence or absence of an injury.

That is, when the fruit table of a fruit or vegetable is imaged over a wide range as described above, it is not easy to uniformly illuminate the entire fruit table to be photographed, so that the amount of light in the peripheral portion is often low. . However, in the apparatus of the present invention, since the saturation of the pixel is compared with the partial average value of the saturation of the surrounding pixels obtained by the moving average method, it is not easy to make the illumination of the entire fruit table uniform. In both cases, the effect is greatly reduced. Specific means for minimizing the influence of fluctuations in light amount, shading, and the like by comparing the central pixel and the peripheral region when determining whether a certain point is a flaw or not is limited. However, for example, ± x in the X direction from a certain point C _n
1 , ± x ₂ (x ₂ > x ₁ ) four points apart or a point (C _nx ) ± x dots away from the point C _{n in the} X direction
), (C _{n + x} ), and two points (C _ny ) and (C _{n + y} ) separated by ± y dots in the Y direction, and the average value of the saturation is calculated. An example is a system.

A plurality of points separated from the center point in a matrix, for example, the above-mentioned two points (C _nx ) and (C _{n + x} ) and ± y dots away from these points in the Y direction. Points (C _n-xy1 ), (C _n-xy2 ), (C _{n + xy1} ), (C
_{n + xy2} ), and a method of calculating an average value of saturation of six points in total. In order to reliably detect a linear flaw, a method of obtaining an average of a plurality of points in the X and Y directions from the center point is preferable. It should be noted that the degree of separation from the center pixel of the peripheral pixels selected for obtaining these partial average values may be appropriately determined within a range of several pixels to several tens of pixels.

As the determination means, the partial average value [ C _n ] of the saturation corresponding to each pixel obtained as described above is used.
And the difference ΔC _n between the original signal C _n and ΔC _n = [C _n ]
−C _n and ΔC _n and a predetermined threshold value C _K
Compared bets can be provided to determine that injury sites exceeding C _K.

In this manner, a flaw can be detected from information in a peripheral range where illumination conditions are substantially uniform, and the influence of variations in the illumination optical system of the entire apparatus can be reduced.

In the automatic injury detection apparatus of the present invention, a luminance (Y) signal can be used in place of the above-described system using the saturation (C) signal for detecting a black system injury. That is, light source means for applying illumination light to fruits and vegetables conveyed a large number apart, light receiving means for imaging the illuminated fruits and vegetables to obtain a luminance signal for each pixel, and a moving average method for each pixel. The apparatus may be configured to include a calculating means for calculating a partial average value of the peripheral luminance, and a determining means for comparing the luminance of each pixel with the corresponding partial average value to determine the presence or absence of an injury-corresponding part.

In this configuration, similarly to the case where the chroma signal is used, the partial average value [ Y _n ] of the luminance around each pixel is obtained by the moving average method, and the difference obtained from this and the original signal Y _n is obtained. By comparing ΔY _n with a predetermined threshold Y _K , it is possible to detect a black-based flaw whose luminance is low.

Further, in the automatic injury detection apparatus for fruits and vegetables according to the present invention, detection exclusion color information setting means for setting detection exclusion color information for a color which is not detected as a flaw, red, and red for each pixel extracted by the light receiving means. A color information calculating means for obtaining color information corresponding to the detection exclusion color information from green and blue signals; and comparing the color information of each pixel obtained by the color information calculation means with the set detection exclusion color information. It is also preferable to provide a detection exclusion determining means for determining whether or not it is necessary to detect a flaw.

For example, when the fruit or vegetable has a settler, the detection exclusion color information setting unit and the red pixel for each pixel taken out by the light receiving unit are provided as means for preventing the settler portion from being erroneously detected as a flaw. , Green and blue signals, and color information calculation means for obtaining color information corresponding to the detection exclusion color information, and detection exclusion determination means for judging whether or not flaw detection is necessary or unnecessary.

In other words, in order to prevent false detection of the filing portion as a flaw, the average color of the fetter given by the type of fruits and vegetables is lightness (Y) 20, hue (H) 160, and saturation (C). In the case of 30, Y = 0 to 40 and H = 1 for each of these three elements, for example, each having an allowable range of ± 20.
40-180, C = 10-50 are set as color information,
As for a pixel that satisfies these three conditions, a pixel having a configuration in which this portion is determined to be a flaw and not a flaw (detection and exclusion) can be exemplified.

In addition, in the case of fruits and vegetables having a mixture of green and yellow fruits, for example, uncolored / semi-ripe mandarin orange, the saturation of the green portion tends to be low, which may cause false detection as a scratch. There is. Therefore, it is preferable to provide a means for preventing the erroneous detection of the green region in the same manner as in the case of the above-mentioned printing.

As means for preventing such erroneous detection of the green area, the conditions of the three elements Y, H, and C of the color to be treated as the green area are used as color information, similarly to the erroneous detection means for the head area. For pixels that satisfy this condition,
An example of a configuration in which this portion is a green region and is determined not to be a scratch (detection is excluded) can be exemplified.

Further, when the prevention of erroneous detection of the head portion or the erroneous detection of the green area is performed for each of a plurality of imaging screens (for example, three screens, five screens, and the like), each screen is provided separately. It is also possible to separately set conditions for the damage detection exclusion color information. By doing so, it is possible to further reduce the influence of shading of an optical system that differs for each screen.

The device of the present invention also sets the detection exclusion color information as described above to prevent false detection of a part such as a settler as a flaw. (For example, brown scars appearing in the fruit table of citrus fruits) are set as specific scar color information by setting the conditions of Y, H, and C in the same manner as described above, and the red, green, and blue colors of each pixel extracted by the light receiving means are set. The color information corresponding to the specific flaw color information is obtained from the signal by the color information calculating means, and the presence / absence of a specific color flaw equivalent site is determined by comparing the set specific flaw color information with the calculated color information. A means for performing this may be provided.

Fruits and vegetables that can automatically detect injuries using the apparatus of the present invention are not limited to specific ones, but oysters, oranges, tomatoes, apples, pears and the like are typical. Can be mentioned.

The method of using the partial average value by the moving average method based on the saturation (or luminance) signal for the detection of injuries according to the present invention is configured as an apparatus which applies this method to the entire surface of a fruit or vegetable image picked up. However, since the illumination is easily performed uniformly, such as the vegetable and vegetable upper surface central area, a portion that can perform injury detection at a high determination level even by the conventional method (the vegetable and vegetable upper surface central area) has a predetermined saturation (or It is also possible to adopt a method of directly comparing the threshold value of the luminance) with the detected chroma signal (or luminance signal), and adopt an injury detection method using the partial average value by the moving average method only for the peripheral region.

In order to divide an area into a central area and a peripheral area, the total luminance or the luminance from any one of the red, green and blue signals is calculated for each pixel. Calculate and compare this with a preset threshold value of luminance (inspection area setting value) to define an area larger than the threshold value as the central area,
It is preferable that an area smaller than the threshold is partitioned as a peripheral area. In addition to this, an imaged pixel group is defined based on a predetermined unit length (or range) within this length (or range). By comparing the magnitude of the amount of change in luminance with a predetermined threshold value and dividing the area into a region having a large amount of change in luminance and a region having a small amount of change, the inspection region can be set as a section.

[0034]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be further described below with reference to the embodiments shown in the drawings.

Embodiment 1 FIG. 1 shows an illumination optical system for picking up a conveyed fruit and vegetables and an image pickup means. In FIG. It is provided so that it may be conveyed. A pair of light sources 3, 3 for irradiating illumination light from diagonally above right and left are arranged in the middle of the conveying path of the conveyor 1, and a light source such as a CCD color camera for imaging the fruits and vegetables 2 illuminated by this light source is provided. An imaging device (R, G, B camera) 4 is disposed right above the conveyor 1.

FIG. 2 shows an illuminating optical system and an image pickup means configured to pick up three sides, that is, both sides and an upper surface of a fruit and vegetable using a single color camera. In FIG. , Fruits and vegetables 2, light source 3, and imaging device 4 are shown in FIG.
Is similar to The feature of this example is that boxes 12 and 13 are arranged on the left and right of the fruits and vegetables 2 to be conveyed, and the reflected light from the fruits and vegetables is reflected by the mirrors 6 and 7 through the slits 16 provided in the diffusion plates 10 and 11. Upper plate 14
Is incident on the imaging device 4 through the slit 15 of the fruit and vegetable 2
Are imaged on both sides and the upper surface of the.

In the images obtained by the illumination optical system and the image pickup apparatus (the three sides of the fruits and vegetables and the upper surface), there are portions where the images overlap each other. For example, the image pickup information is stored in a memory (not shown). Then, when this is read, by selecting and reading the central portion of each image, it is possible to obtain imaging information of a continuous surface with little shading.

The fruits and vegetables having (a) brown scratches, (b) white scratches, (c) brown scratches + white scratches, and (d) scabs in FIG. Signal processing for flaw detection performed in the case of imaging will be described with reference to FIG. 4 showing the signal processing circuit. FIG. 3 shows a case where the fruits and vegetables are imaged from the side for convenience of explanation.

The image pickup device (R, G,
The red (R), green (G), and blue (B) signals of each pixel captured by the (B camera) 4 are A / D converted by an A / D converter 20.
After the conversion, it is input to the image processing computer, and the three elements of color, brightness (Y), hue (H), and saturation (C), are respectively determined according to a known predetermined arithmetic expression set in the color converter 21. It is calculated for each pixel and stored in a memory (not shown).

Separately, the total brightness T is calculated by the T calculator 40 based on the RGB signals. This calculation of the total luminance T is for dividing the inspection area of the fruit table into a central area and a peripheral area or an upper area and a lower area. In this example, the comparator 42 gives the inspection area to the inspection area setting means 41. By comparing the brightness with the threshold value, a region larger than the threshold value is detected as the center region or the upper region, and a region smaller than the threshold value is detected as the peripheral region or the lower region. Is output to

The three color components Y, H, and C are read from a memory (not shown) in which the three components are stored. In this example, three signal processes are performed in parallel.

The first of which the parallel signal processing is performed is the determination of the presence or absence of an injury-corresponding site based on the saturation C.
That is, in this example, the partial average calculator 22, partial average value of saturation corresponding to a pixel C _n [C _n] is, for each pixel, a predetermined number of pixels in the two directions perpendicular to that pixel as a center The difference ΔC _n between the partial average value [ C _n ] obtained by the difference detector 23 and the saturation C _n of each pixel is obtained by averaging the saturation of the pixels at a plurality of points that are apart from each other. when exceeding the predetermined threshold value C _K set in detection sensitivity setting unit 24, comparator 25 is provided so as to determine the wound site, and outputs. In this example, the AND circuit 2
By taking the logical product of the comparator 42 and the above-described comparator 42, if the area is a region where the flaw detection based on the saturation is performed (for example, only the peripheral area of the fruit table or the whole area of the fruit table), the defect of the chroma difference The output is as it is.

The second of the parallel signal processing is to prevent false detection of a print area. In this example, the print color information setting means 27 uses, for example, Y = 0 to 40 and H =
140 to 180 and C = 10 to 50 are set as the fly color information. Then, corresponding Y, H, C for each pixel
And the set letter color information are compared by the comparator 28, and a pixel having the color of the letter part is detected. The scab part detected in this way is usually near the upper center of the fruit or vegetable. In this example, in order to prevent erroneous detection in the vicinity of the area detected as the head part, an outer peripheral area for a certain number of pixels is set by the enlargement calculator 29 outside the group of pixels detected as the head part. However, the inside of the outer region is determined to be a fly region. Then, for the pixel determined to be the print area, the above-described enlargement calculator 2
9 is inverted by an inverter 30 and is ANDed with an output from an AND circuit 36 (to be described later) (AND circuit 31). If it is a part, it is configured not to be detected as a scratch.

Thirdly, in order to detect a specific color flaw, the numerical range of Y, H, and C corresponding to the color flaw is previously determined as specific flaw color information in order to detect a specific color flaw. The comparator 33 compares the Y, H, and C signals for each pixel set in the color information setting means 32 and read from a memory (not shown) with the flaw color information set in the flaw color information setting means 32. As a result, a pixel including any of them in the set numerical value ranges of Y, H, and C is determined to be a specific color flaw, and the AND circuit 35 performs a logical product with the comparator 42. Thus, in the case where the flaw detection is performed (for example, only the center area of the fruit table, only the peripheral area of the fruit table, or the entire area of the fruit table), an output indicating that there is a color flaw is output.

In the circuit of this embodiment, the OR of the outputs of the AND circuits 26 and 35 is output by the OR circuit 36 for each pixel. 31 is established, and the flaw signal of the corresponding pixel is finally output.

FIGS. 5 to 8 show signal outputs of the brightness (Y), hue (H), and saturation (C) obtained by the color converter 21 in the signal processing circuit described in FIG. FIGS. 3A to 3C show states, and these figures are for convenience sake.
Y, H, and C detected as a single linear pixel row in the left-right direction of the drawing so as to include a portion such as a scratch for each of (d).
Showed a change.

From these figures, the detected values of the Y, H, and C signals are higher from the conveyor surface (the left side of the horizontal axis in the figure) to the upper surface of the fruits and vegetables (the right side of the horizontal axis in the figure). In addition, the saturation signal C is characteristically changed at the injured part or the spot area (see FIGS. 5 to 8), and the hue H signal is characteristically increased at the spot area (see FIG. 8).
I understand.

FIG. 9 shows the Y, H, and C signals collectively by changing the scale of the horizontal axis in FIGS. 5 to 8, and sequentially shows FIGS. 5 and 6 from the left side of FIG. , FIGS. 7 and 8 respectively show the results.

The comparator 2 of FIG. 4 corresponding to FIG.
FIG. 1 shows the output signal of FIG. 5 (the relative saturation change by the moving average method).
0. Note that the partial average value in the present example was obtained as a moving average of a point three pixels (dots) left and right from the center pixel.

As can be seen from this figure, the parts of the fruits and vegetables of FIG. 3 where the saturation is lower than the surroundings are shown in FIG.
It can be seen that the detection can be clearly performed at 0. Note that, in this example, in the case of FIG.
Although the output of No. 5 is detected as a flaw, the inverter 30 does not output this flaw portion (the detection output is inverted), so that no flaw signal is output from the AND circuit 31.

Example 2 This example is directed to an example in which oranges whose fruit table is almost entirely yellow and some green parts are present are excluded, and the green parts are excluded from being detected as scratches. ing.

That is, as shown in FIG.
If the whole surface of the fruit table is yellow and there is a green part in the part, the saturation of this green part is low and tends to be detected. For this reason, when the relative change of the saturation by the moving average method in FIG.
The curve of the partial average value (shown by the broken line) obtained by the moving average method shows a higher value than the curve of the original signal of the saturation (shown by the solid line) in the middle graph of the figure. As shown, the output of the comparator 25 in FIG. 4 may be detected as a flaw.

Therefore, in the present embodiment, a signal processing circuit (not shown) for detecting and excluding the green portion is configured with the same configuration as that for detecting and excluding the print portion in FIG. The setting of the detection exclusion color information for excluding the detection of green may be a method of setting the condition range of the Y, H, and C signals similarly to the setting of the color information of the print area, or may be a method of setting the green hue. Alternatively, a method may be used in which a green hue signal corresponding to the captured R, G, B signals is obtained and compared.

[0054]

According to the present invention, the detection accuracy is greatly improved as compared with the conventional apparatus for mechanically detecting an injury, and particularly, the temporal change in the illumination and imaging optical systems,
Even if there is a fluctuation, an effect is obtained that the damage can be detected stably without substantially affecting the detection accuracy.

Further, the influence of shading in the illumination of fruits and vegetables is reduced, and even when the light amount in the peripheral region is likely to be lower than that in the central region, each pixel is compared by the moving average method with the peripheral region as the background. The advantage is that high detection accuracy can be maintained.

Further, the possibility of erroneously detecting a green portion of a fruit or vegetable with a letter or a tangerine as a flaw can be surely prevented by performing signal processing for excluding it from being detected as a flaw as necessary. Therefore, there is an effect that the possibility of erroneous detection is greatly reduced.

Further, depending on the fruits and vegetables, it may be required to detect a characteristic color damage, and in this case, the characteristic color damage can be detected.
The effect of improving the accuracy of flaw detection is obtained.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of an arrangement relationship between an illumination optical system and an imaging device used for implementing the present invention;

FIG. 2 is a diagram showing another example of an arrangement relationship between an illumination optical system and an imaging device used in the embodiment of the present invention;

FIGS. 3A and 3B are diagrams showing examples of wounds and the like of fruits and vegetables to be subjected to wound detection by the apparatus of the present invention, wherein FIG. 3A shows fruits and vegetables with brown scratches, FIG. 3B shows fruits and vegetables with white scratches, and FIG. ) Shows fruits and vegetables with both brown and white wounds, and (d) shows fruits and vegetables with scabs.

FIG. 4 is a block diagram showing an outline of a configuration of a circuit for performing signal processing of the device of the present invention;

FIG. 5 is a diagram showing a change curve of Y, H, and C signals of one line of linear pixels obtained for the fruits and vegetables of FIG. 3A;

FIG. 6 is a diagram showing a change curve of Y, H, and C signals of one line of linear pixels obtained for the fruits and vegetables of FIG. 3B;

FIG. 7 is a diagram showing a change curve of Y, H, and C signals of one line of linear pixels obtained for the fruits and vegetables of FIG. 3C;

FIG. 8 is a diagram showing a change curve of Y, H, and C signals of one line of linear pixels obtained for the fruits and vegetables of FIG. 3D;

FIG. 9 is a diagram showing the curves of FIGS. 5 to 8 collectively by changing the horizontal axis scale,

FIG. 10 is a diagram illustrating a relative saturation change obtained by a moving average method based on the saturation signal of FIG. 9;

FIG. 11 is a view for explaining detection and exclusion of a green portion in oranges;

[Explanation of symbols]

1 ... conveyor, 2 ... fruits and vegetables, 3 ... light source, 4
... Imaging device, 6,7 ... Mirror, 10,11 ...
Diffusion plate, 12, 13 ... Box, 14 ... Upper plate, 15, 16 ... Slit, 20 ... A / D converter, 21 ... Color converter, 22 ... Partial average value calculator, 23: difference detector, 24: flaw detection sensitivity setting means, 25: comparator, 26: AND circuit, 27
... Sheet color information setting means, 28 ... Comparator, 29
..Enlargement arithmetic unit, 30 ... inverter, 31 ... AND circuit, 32 ... scratch color information setting means, 33 ... comparator, 35 ... AND circuit, 36 ... OR circuit , 40
... T calculator, 41 ... inspection area setting means, 42
..Comparators.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G01N 21/85 G01N 21/892 G06T ⁷ /00-7/60

Claims (7)

(57) [Claims] 1. Light source means for applying illumination light to fruits and vegetables conveyed at a large distance, light receiving means for taking an image of the illuminated fruits and vegetables and extracting red, green and blue signals for each pixel; Calculating means for calculating the saturation (C) of each pixel based on the red, green, and blue signals of each pixel, and calculating a partial average value of the saturation around each pixel by a moving average method; Judging means for judging the presence or absence of an injury-equivalent part by comparing the saturation with the partial average value corresponding thereto; 2. Light source means for applying illumination light to fruits and vegetables conveyed a large number apart, light receiving means for taking an image of the illuminated fruits and vegetables and obtaining a luminance signal for each pixel, and each pixel by a moving average method. Calculation means for calculating a partial average value of the peripheral luminance for each pixel, and determining means for comparing the luminance of each pixel with the corresponding partial average value to determine the presence or absence of an injury-corresponding part. Automatic detection device for fruits and vegetables. 3. The partial average value according to the moving average method according to claim 1 or 2 is obtained as an average value of pixels at a plurality of points separated by a predetermined interval in two directions orthogonal to each pixel. An apparatus for automatically detecting injury of a vegetable or vegetable, wherein the average value is obtained as an average value of pixels at a plurality of points separated by a predetermined distance in a matrix with each pixel as a center. 4. An apparatus for automatically detecting injuries of fruits and vegetables according to claim 1, wherein said detection exclusion color information setting means for setting detection exclusion color information for a color not detected as a flaw; A color information calculating means for obtaining color information corresponding to the detection exclusion color information from green and blue signals; and comparing the color information of each pixel obtained by the color information calculation means with the set detection exclusion color information. Key to detection as scratches
An automatic injury detection apparatus for fruits and vegetables, comprising: a detection exclusion determination means for determining unnecessaryness. 5. The automatic injury detection apparatus for vegetables and fruits according to claim 4, wherein when the detection exclusion color information to be set is a fruit and vegetables with a letter, the information is color information of the letter. 6. The automatic vegetable injury detection apparatus according to claim 4, wherein the detection exclusion color information set is green when the fruits and vegetables are citrus. 7. A flaw / injury automatic detection apparatus according to claim 1, further comprising: flaw color information setting means for setting flaw color information for a specific color flaw;
Color information calculating means for obtaining color information corresponding to the flaw color information from green and blue signals; and comparing the color information of each pixel obtained by the color information calculation means with the set flaw color information to specify the color information. And a judging means for judging the presence / absence of a site corresponding to a color wound.