Specific implementation mode
The invention will be further described with the following Examples.
Referring to Fig. 1, a kind of intelligent orchard management system of the present embodiment, plantation apple in orchard, including environmental monitoring
Sensor 1, signal transmitting apparatus 2 and managing device 3, the signal transmitting apparatus 2 are established environmental monitoring by mobile network and are passed
Communication between sensor 1 and managing device 3, the environment monitoring sensor 1 are used to monitor the environmental parameter in apple orchard, and will
The environmental parameter is sent to managing device 3 by signal transmitting apparatus 2, and the managing device 3 includes irrigation rig and picking
Device, the irrigation rig are used to be irrigated according to environmental parameter, and the picker is for picking transparent apple.
A kind of intelligent orchard management system is present embodiments provided, intelligence pipe of the orchard from irrigating picking is realized
Reason.
Preferably, the picker includes obstacle detector and apple-picking device, the detection of obstacles dress
It sets and is detected for the barrier to apple orchard, the apple-picking device is for picking transparent apple;The barrier
Hinder analyte detection device for being detected to the barrier in apple orchard:The ambient image for acquiring apple orchard, by the image in apple orchard
Lab color spaces are transformed into, barrier is detected according to red green component a in Lab color spaces, set detection threshold value EH,
EH ∈ [- 20, -25], if the red green component a < EH of pixel, as background parts, if the red green component a >=EH of pixel,
Then as barrier.
This preferred embodiment picker is considered as background during detection of obstacles, by green portion, and remaining is regarded
Make barrier, meet the environmental characteristic in apple orchard, realize quick detection of obstacles, to obstacle during apple-picking
Object is hidden.
Preferably, the apple-picking device includes that collecting unit, detection unit, cutting unit, filter unit, feature carry
Unit, recognition unit and manipulator are taken, the collecting unit is for acquiring tree crown Apple image, and the detection unit will be for that will set
Hat Apple image is transformed into Lab color spaces, is detected to apple, and the cutting unit is used for transformed tree crown apple
Image is split, and for being filtered to the image after segmentation, the feature extraction unit is used for the filter unit
The feature of apple is extracted according to the image after being filtered, the recognition unit is identified apple according to feature,
The manipulator picks apple according to recognition result.
In image acquisition process, since branch, trunk, leaf block, in the image of acquisition apple mostly loses carefully
Contour feature is saved, object edge situation is also remote undesirable, and prodigious difficulty is brought for tree crown apple identification, due to planting garden
The contextual factors such as sky, illumination, house, shade influence, the fruit shape feature of apple also tends to be submerged in all kinds of backgrounds or image
In noise.This preferred embodiment apple-picking device realizes apple by converting unit, the processing of cutting unit, filter unit
The accurate extraction of fruit feature is laid a good foundation for accurately identifying for tree crown apple.
Preferably, the detection unit is used to tree crown Apple image being transformed into Lab color spaces, is examined to apple
It surveys:Apple is detected according to red green component a in Lab color spaces, setting detection threshold value EM, EM ∈ [- 5, -6.5], if picture
The red green component a < EM of vegetarian refreshments, then as background parts, if the red green component a >=EM of pixel, as apple can be picked
Fruit;
The process of ripe apples is regarded as by green to red process, this meets the spy of red green component in Lab color spaces very much
Point, this preferred embodiment detection unit make full use of this promise that can pick the accurate detection of apple, pass through adjusting detection threshold
Value, can adjust the maturity that can pick apple, and detection threshold value is bigger, indicate that the maturity that can pick apple is higher.
Preferably, the cutting unit includes gray processing unit and Threshold segmentation unit, and the gray processing unit is used for will
Red green component image is converted into gray level image in Lab color spaces, and the Threshold segmentation unit is used to carry out threshold to gray level image
Value segmentation, apple is partitioned into from complicated background;The Threshold segmentation unit includes the first cutting unit, the second cutting unit
With fusion cutting unit, first cutting unit is for obtaining first object region, and second cutting unit is for obtaining
Second target area, the fusion cutting unit are used to determine target according to first object region and the second target area.
First cutting unit is for obtaining first object region:If size is the gray level image W of M × N, number of greyscale levels
Gray value for L, pixel (m, n) is hm, n), m ∈ { 1,2 ..., M }, n ∈ { 1,2 ..., N }, 0≤h (m, n)≤L-1, profit
The first segmentation function of image is determined with following formula:In formula, CSW
Indicate the first segmentation function of image, i=h (m, n), piIndicate that gray value is that the pixel of h (m, n) accounts for total pixel number amount in image
Number, t indicate the first segmentation threshold;The first segmentation function is maximized, best first segmentation threshold t ' is obtained: It will be greater than the pixel of best first segmentation threshold
Point is used as first object region, is less than the pixel of best first segmentation threshold as background.
Second cutting unit is for obtaining the second target area:Calculate 5 × 5 neighbours centered on pixel (m, n)
The average gray value k (m, n) in domain, 0≤k (m, n)≤L-1 determine the second segmentation function of image using following formula: In formula, ZCWIndicate the second segmentation of image
Function, j=k (m, n), p (i, j) indicate that grey scale pixel value is i in image and the average gray value of 5 × 5 neighborhood of pixel is j's
The number of pixel accounts for the ratio of the total pixel of image, and (r, s) indicates the second segmentation threshold;The second segmentation function is maximized, is obtained
To best second segmentation threshold (r ', s '):It will be big
In best second segmentation threshold pixel as the second target area, be less than the pixel of best second segmentation threshold as the back of the body
Scape.
The fusion cutting unit is used to determine target according to first object region and the second target area:By first object
The intersection of region and the second target area is as target, i.e. tree crown apple.
Under normal circumstances, it is converted into the image after gray scale and shows have no the characteristics of rule.To from the irregular ash of milli
It is adaptive in degree image to isolate picking target, it is necessary to the reasonable effective threshold segmentation method of selection.This preferred embodiment point
The method that unit uses twice threshold segmentation is cut, target, i.e. tree crown apple is accurately determined.
Orchard is managed using intelligent orchard management system of the invention, 5 apple orchards is chosen and is tested, respectively
For apple orchard 1, apple orchard 2, apple orchard 3, apple orchard 4, apple orchard 5, management cost and the efficiency of management are counted, with artificial
Management is compared, and generation has the beneficial effect that shown in table:
|
The efficiency of management improves |
Management cost reduces |
Apple orchard 1 |
29% |
27% |
Apple orchard 2 |
27% |
26% |
Apple orchard 3 |
26% |
26% |
Apple orchard 4 |
25% |
24% |
Apple orchard 5 |
24% |
22% |
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than the present invention is protected
The limitation for protecting range, although being explained in detail to the present invention with reference to preferred embodiment, the ordinary skill apple orchard of this field
It should be appreciated that can be modified or replaced equivalently to technical scheme of the present invention, without departing from technical solution of the present invention
Spirit and scope.