CN102737367A - Tea image enhancement and division method based on color characteristic - Google Patents

Abstract

The invention relates to the field of an imaging processing technology and discloses a tea image enhancement and division method based on a color characteristic. The tea image enhancement and division method comprises the following steps of: firstly, converting an original tea RGB (Red, Green and Blue) colored image into an HIS (Hue, Intensity and Saturation) color space; carrying out convergence calculation based on a characteristic hue on a hue H and carrying out adjustment and calculation based on an expected value on a saturation S; then, carrying out primary seed selection on hue and saturation parameters, and then carrying out regional growth on the seed region based on the similarity of colors and the connective characteristics of the region; combining a color distance and an edge distance to carry out the growth and the combination of the region; and finally, finishing the division of tea tender shoots. With the adoption of conversion, enhancement and division tests of a tea image color space, a result shows that the algorithm can better keep the color characteristic and can better divide the tea tender shoots from a tea image, so that a new method is provided for growth monitoring and nondestructive testing for teas.

Description

Tealeaves figure image intensifying and dividing method based on color characteristic

Technical field

The invention belongs to technical field of image processing, be a kind of to on-the-spot tealeaves image based on the tealeaves figure image intensifying of color characteristic and the method cut apart based on the tealeaves tender shoots that region growing merges.

Background technology

Image processing techniques be along with the development of computing machine with ripe and developing rapidly in recent years-individual important application technical field; In modern times in the production of precision agriculture; Utilize computer image technology can greatly improve precision and efficient; Realize the robotization and the intellectuality of information acquisition and processing, and can can't harm crop growth, fast, real-time monitoring.Applied Computer Techniques is to analysis and the more existing researchs of identification of tea color and quality; But mostly research in the past is based on studying to the shape of finished tea and color or to the soup look of tealeaves, to the rare both at home and abroad corresponding research that is segmented in of the tealeaves tender shoots in the on-the-spot tea tree in farmland.

Growth conditions of tealeaves tender shoots and degree of ripeness play an important role to the harvesting of tealeaves and the tealeaves end product quality in later stage; The growth conditions of discerning and analyze the tealeaves tender shoots in time has positive effect to the standardization large-scale production of tealeaves, and is the analysis and the base of recognition of carrying out tealeaves to cutting apart of tealeaves tender shoots image.

Color space is the theoretic characteristic manner of coloured image different colours; Original digital image is a rgb format; The present invention at first with the tealeaves image from the RGB color space conversion to the HSI color space; The HSI color space more can show our understanding to color near the view mode of people to the colored world better, and it representes color through tone H (Hue), saturation degree S (Saturation), three attributes of brightness I (Intensity); In the HSI space; Correlativity between H, S, the I three-component is than much little between R, G, the B three-component, the advantage in HSI space be it with brightness (I) and reflection color intrinsic propesties two parameters---tone (H) and saturation degree (S) are separated, and make Flame Image Process still less receive the influence of illumination condition.It is full as to spend the expression to image also more to help color harmony, therefore based on the tealeaves image Segmentation in HSI space, can obtain than the better effect of rgb space.

Image segmentation is meant with the zone of satisfying specified conditions that Special Significance is arranged in the image with other zone separately, is mutually disjointed mutually in these zones.Existing color image segmentation method is existing multiple, and threshold method commonly used requires in the histogram obvious Wave crest and wave trough is arranged, and in the processing of coloured image, causes easily to cut apart coarsely, and performance is bad; And the feature space split plot design has merged the local message and the spatial information of image area, and local decision-making noise resistance interference is strong, but is prone to cause less divided.The present invention adopts region growing act of union effectively Fusion of Color information and spatial information, and can comprehensive local message and global information, can be simultaneously non-conterminous but zone color similarity be spatially merged respectively.

Summary of the invention

The present invention concerns with similarity in conjunction with the characteristic between each zone in tealeaves tender shoots and the tea tree ' s old leaf image is different through analyzing the color space characteristic of tea tree image.A kind of tealeaves figure image intensifying and dividing method based on color characteristic proposed; At first the tealeaves original image is carried out color space conversion; And the color harmony saturation degree in the HSI space after the conversion is carried out the figure image intensifying calculate under the condition more outstanding based on color characteristic; Original unclear image is become clear and stresses tealeaves tender shoots characteristic out, obtain visual effect better, more useful image.By improved new algorithm, combine the growth in color and zone and merge the tealeaves tender shoots is cut apart simultaneously based on region growing.The method has combined the color characteristic of tealeaves image and the global information and the local message of provincial characteristics preferably, can accomplish a plurality of tealeaves tender shoots in the tealeaves image better to cut apart, and obtain good experiment effect.

Tealeaves figure image intensifying and dividing method based on color characteristic of the present invention is characterized in that comprising following concrete steps:

(1) obtains tealeaves original image in the tea tree;

(2) with original image from the RGB color space conversion to the HSI color space, and choose H and S parameter in the model as the characteristic of tealeaves image;

(3) tone H is carried out converging calculating based on the figure image intensifying of characteristic tone;

(4) saturation degree S is carried out calculating based on the figure image intensifying adjustment of expectation value;

(5) in image, select the Partial Feature pixel as seed;

(6) based on the growth rule seed region is grown, will be attached on the seed of growth district with the neighbor of tealeaves tender shoots seed color similar performance;

(7) scan and choose based on merging many sub-block of rule to entire image, to close on color, adjacent areas merges on the space;

(8) expand and the morphology that shrinks is handled in the zone after being combined;

(9) accomplish cutting apart of a plurality of tealeaves tender shoots.

H span in the HSI color space is [0,360], and the span of S and I is [0,1], the RGB tlv triple of each pixel in the tealeaves image (R, G, B) a corresponding HSI tlv triple (H, S, I).H is 0 and represented that redness, the color of intermediate value then were that order changes successively at 360 o'clock; S represented that saturation degree was minimum at 0 o'clock, was to represent that saturation degree was the highest at 1 o'clock.

In the tealeaves image, the tealeaves tender shoots has outstanding peak green, and the tone images Enhancement Method that the present invention proposes is: defined two tealeaves tender shoots characteristic TINTCs; On the hue circle plane, experiment finds that these two tone value correspondences are respectively 130 and at 160 o'clock, can obtain effect preferably; Be the center with these two values simultaneously; Near the central value that satisfies condition tone value is converged to central value by various conditions and step-length,, make each tone value more approach the center tone to reduce the number of less important relatively tone; Reach and make the stronger purpose of original image gradation sense, also be more conducive to cutting apart of next step tealeaves tender shoots simultaneously.

Step-length is very important to choosing of converging, and step-length is crossed conference and caused image ill effects such as color change to occur, and the too small meeting of step-length makes regulating effect not obvious.Through experiment test repeatedly, confirm the step-length that converges that step-length elects 1,2 as, 3 three kind is different, little near the step-length of TINTC, the step-length away from TINTC is big more more.The calculating of converging of tone is defined as:

H _i′＝H _i±k

1＜|H _i-H ₀|≤5 k＝±1

Wherein, 5＜| H _i-H ₀|≤10 k=± 2

10＜|H _i-H ₀|≤15 k＝±3

H _iBe the tone value of certain point in the image, H _i' converge the tone value after the calculating, H ₀Be TINTC, k is a step-length, works as H _i＜H ₀The time, the k value is worked as H for just _i＞H ₀The time, the k value is for negative.

And saturation degree is another factor that influences coloured image.It is distincter and pure that image is had, but do not consider the saturation degree characteristic that image itself is original, then may make the too high and whole structure of destruction original image of image saturation, ill effects such as generation supersaturation, cross-color.The present invention adopts statistical method to draw the saturation distribution characteristic of original image, and rationally regulates on this basis.

Mathematical expectation is an important numbers characteristic of probability distribution.The mathematical expectation of stochastic variable is exactly the weighted mean of stochastic variable.In processing to image saturation; The probability of at first calculating each pixel saturation value in the image and in image, occurring; Obtain its mathematical expectation E (x);

also adjusts the saturation degree of image with this, and the adjustment formula definition is:

S _iBe original saturation degree component, S _i' be adjustment back saturation degree component, α is a stretching factor, the degree of saturation of decision saturation degree component, x _iBe the value of saturation degree in the image, p _iBe the corresponding probability that occurs of this intensity value.

In to the growth course of seed region in order to show that there is very high similarity in a zone and its neighborhood zone; Be similarity function of each zone definitions; Weigh the similarity degree in the zone of a zone and its neighborhood with similarity function; If the similarity function value in a zone greater than a threshold value, then should just can be selected as seed region in the zone.

With a region R _{I '}The neighborhood sets definition be:

$U_{i^{\′}}^R=R_{i^{\′}}\∪R_{j^{\′}}/R_{j^{\′}}\∈R$

J '=1,2 wherein ... k ', R are the entire image collection, R _{I '}Be some zones wherein, R _{J '}Be its neighboring region, the similarity function in the zone of a zone and its neighborhood be defined as:

$f(R_{i^{\′}},U_{i^{\′}}^R)={\mathrm{\ϵ}}_1\·\mathrm{simH}\left(R_{i^{\′}}\right)+{\mathrm{\ϵ}}_2\·\mathrm{simS}\left(R_{i^{\′}}\right)$

Wherein $\mathrm{SimH}\left(R_{i^{\′}}\right)=\sqrt{\frac{{\mathrm{\Σ}}_{i^{\′}=1}^{k^{\′}+1}{(x_{i^{\′}}-\stackrel{\‾}{x})}^2}{k^{\′}+1}}$

x _{I '}For

In each regional tone H component average,

For

The tone component average of All Ranges in the set, k ' is an integer;

And $\mathrm{SimS}\left(R_{i^{\′}}\right)=\sqrt{\frac{{\mathrm{\Σ}}_{i^{\′}=1}^{k^{\′}+1}{(y_{i^{\′}}-\stackrel{\‾}{y})}^2}{k^{\′}+1}}$

y _{I '}For

In each regional saturation degree S component average, For

The saturation degree component average of All Ranges in the set.And ε ₁And ε ₂Be the tone component and the saturation degree component coefficient of definition.

In the zone of image merges, think that two zones are close on color, adjacent on the space, and not have significant edge be two zones that can link to each other at its neighborhood place.Promptly the maximal value of the relative color distance in a zone and its neighborhood zone is less than the threshold value of definition.Adopt the tone component average define color distance in zone to calculate, define as follows:

$D_c=\frac{r_i\·r_j}{r_i+r_j}\left|\right|{\stackrel{\‾}{\mathrm{\μ}}}_i-{\stackrel{\‾}{\mathrm{\μ}}}_j\left|\right|$

R wherein _iAnd r _jRepresent the number of pixels that comprises in i and the j zone respectively, With

Represent the tone average in two zones, || || the expression Euclidean distance.r _iR _jProduct make comprise the less zone of number of pixels compare with other regional color distances less, thereby under the identical situation of tone average, help the preferential merging of zonule, make segmentation result meet people's visual characteristic more.For initial cut zone; The stepped zone merge algorithm that adopts is accomplished last cutting apart; Be each combined diagram as two the most close zones of middle distance, form a new sub-piece zone, new region is also adjusted with other regional neighborhood relationships and distance thereupon simultaneously.

Choose in the process simultaneously in sub-piece zone to the tender shoots of tealeaves; Might be because close little zone be also arranged on old blade, make and select a plurality of seed regions, the merging that makes the mistake in the segmentation area; In order to address this problem; Some restrictions are carried out in the merging in zone, bigger difference is arranged, come effectively to distinguish through Edge Distance in conjunction with shape size between tealeaves tender shoots and the Lao Ye.The definition Edge Distance is:

$D_e=\frac{1}{P_{\mathrm{ij}}}\underset{m,n}{\mathrm{\Σ}}\left|\right|x_m-x_n\left|\right|$

P wherein _IjNumber of pixels on the expression edges of regions, x _mAnd x _nRepresent the m of both sides of edges and the position coordinate value that n is ordered respectively.|| || the expression Euclidean distance.Require simultaneously:

(1) the equal value difference of tone and the equal value difference of saturation degree that have only two adjacent areas could merge these two zones all less than certain threshold value so.

(2) if the number of pixels in a zone surpasses certain limit less than certain threshold value and Edge Distance, so with this zone merge to the minimum neighborhood zone of its tone average difference in go.

Description of drawings

Fig. 1 is the enhancing and the partitioning algorithm process flow diagram of tealeaves image.

Fig. 2 is the original image of tealeaves.

Fig. 3 is an image after the tealeaves enhancement process.

Fig. 4 is the segmentation result of tealeaves tender shoots

Embodiment

The digital camera that experiment is used is CANON S80, in image capture process, adopts the close shot pattern; Close flashlamp; To avoid of the influence of flashlamp self light to the tealeaves color, should under natural light, carry out capture simultaneously, avoid the direct projection of sunlight; The imaging focal length of getting is 25cm in capture, and resolution adopts 1600 * 1200.

Experiment finds that when the capture focal length was nearer, the tealeaves tender shoots of being obtained was less, and the accuracy rate of cutting apart is higher, and the accuracy rate of when 4-8 tender shoots, cutting apart is the highest.When focal length is nearer, can influence the sharpness of the depth of field and imaging, cut apart accuracy thereby influence, but when focal length farthest the time, the tender shoots image reduces, the scope of image color and luster broadens, and the accuracy rate of cutting apart also can descend to some extent.

Merging at seed region is through each seed region neighborhood of searching loop; Check the neighborhood window of each seed region line by line; Relatively if the tone average in seed region and neighborhood zone and the difference of saturation degree average less than certain threshold value, are then carried out the zone and are merged.Circulation is carried out, till the difference of tone average between any two zones and saturation degree average is all no longer less than preset threshold.Before circulation each time, upgrade the tone average and the saturation degree average of new region simultaneously, remove merged zone in the original neighborhood zone, each zone, keep up-to-date neighborhood relationships between each zone and other zones.Tone average difference limen value is made as 0.05 in the experiment, and saturation degree average difference limen value is made as 0.065.In testing simultaneously, ε ₁Value 0.75, ε ₂Value 0.25

The tealeaves tender shoots is having bigger difference on the characteristic of color and between original Lao Ye, so color characteristic is an effective criterion of cutting apart, because in the image acquisition procedures of tea tree scene to tealeaves; It is bigger influenced by extraneous light; Original picture format is a rgb format, but between the R, G, B three-component very strong correlativity is arranged, with the variation of illumination condition; R, G, three components of B all can have bigger variation; Directly utilize these components often can not obtain required effect, so, choose the HSI space in the choosing of image color space.

Under the HSI model, the color information of image is mainly reflected by H and S, and is following to the conversion formula in HSI space from RGB:

$H=\left\{\begin{array}{cc}2\mathrm{\π}-\mathrm{\θ}& B>G\\ \mathrm{\θ}& B\≤G\end{array}\right.$

Wherein $\mathrm{\θ}=\mathrm{Arccos}\left[\frac{[(R-G)+(R-B)]/2}{\sqrt{{(R-G)}^2+(R-B)(G-B)}}\right]$

$S=1-\frac{3\min (R,G,B)}{R+G+B}$

$I=\frac{1}{3}(R+G+B)$

Application this method reaches figure image intensifying and the split-plot experiment of tealeaves tender shoots to the color harmony saturation degree to the conversion in tealeaves color of image space; The result shows that this algorithm can keep color characteristic and well the tealeaves tender shoots being split well from the tealeaves image, for the growth monitoring and the Non-Destructive Testing of tealeaves provides new method.

Claims (6)

1. tealeaves figure image intensifying and dividing method based on a color characteristic is characterized in that comprising following concrete steps:

(1) obtains tealeaves original image in the tea tree;

(2) with original image from the RGB color space conversion to the HSI color space, and choose H and S parameter in the model as the characteristic of tealeaves image;

(3) tone H is carried out converging calculating based on the figure image intensifying of characteristic tone;

(4) saturation degree S is carried out calculating based on the figure image intensifying adjustment of expectation value;

(5) in image, select the Partial Feature pixel as seed;

(6) based on the growth rule seed region is grown, will be attached on the seed of growth district with the neighbor of tealeaves tender shoots seed color similar performance;

(7) scan and choose based on merging many sub-block of rule to entire image, to close on color, adjacent areas merges on the space;

(8) expand and the morphology that shrinks is handled in the zone after being combined;

(9) accomplish cutting apart of a plurality of tealeaves tender shoots.

2. tealeaves figure image intensifying and dividing method based on color characteristic according to claim 1; It is characterized in that: tone H is carried out converging in the calculating based on the figure image intensifying of characteristic tone in step (3); Two tealeaves tender shoots characteristic TINTCs have been defined; Near the central value that satisfies condition tone value is converged to central value by various conditions and step-length, and its step-length elects 1,2 as, 3 three kind different converges step-length, little near the step-length of TINTC; Step-length away from TINTC is big more more, and calculating is defined as:

Wherein,

H<sub >i</sub>Be the tone value of certain point in the image,<img file="DEST_PATH_FDA00001902056000013.GIF" he="59" id="ifm0003" img-content="drawing" img-format="GIF" inline="no" orientation="portrait" wi="62" />Converge the tone value after the calculating, H<sub >0</sub>Be TINTC, k is a step-length, works as H<sub >i</sub><h<sub >0</sub>The time, the k value is worked as H for just<sub >i</sub>>H<sub >0</sub>The time, the k value is for negative.

3. tealeaves figure image intensifying and dividing method based on color characteristic according to claim 1; It is characterized in that: saturation degree S is carried out in the figure image intensifying adjustment calculating based on expectation value in step (4); The probability of at first calculating each pixel saturation value in the image and in image, occurring; Obtain its mathematical expectation E (x); also adjusts the saturation degree of image with this, and the adjustment formula definition does

S _iBe original saturation degree component, Be adjustment back saturation degree component, α is a stretching factor, the degree of saturation of decision saturation degree component, x _iBe the value of saturation degree in the image, p _iBe the corresponding probability that occurs of this intensity value.

4. tealeaves figure image intensifying and dividing method based on color characteristic according to claim 1; It is characterized in that: in the seed region growth of step (6),, be similarity function of each zone definitions in order to show that there is very high similarity in a zone and its neighborhood zone:

Wherein

x _{I '}For

In each regional tone H component average,

For

The tone component average of All Ranges in the set, k ' is an integer; And ε ₁And ε ₂Be the tone component and the saturation degree component coefficient of definition, the neighborhood set

Wherein R is the entire image collection, R _{I '}Be some zones wherein, R _{J '}Be its neighboring region, j '=1,2 ... k ';

And

y _{I '}For

In each regional saturation degree S component average,

For

The saturation degree component average of All Ranges in the set.

5. tealeaves figure image intensifying and dividing method based on color characteristic according to claim 1; It is characterized in that: two zones are close on color in step (7); Adjacent on the space, and its neighborhood place does not have significant edge then can think two zones that can link to each other; Adopt the tone component average define color distance in zone to calculate, define as follows:

R wherein _iAnd r _jRepresent the number of pixels that comprises in i and the j zone respectively,

With

Represent the tone average in two zones, ‖ ‖ representes Euclidean distance.

6. tealeaves figure image intensifying and dividing method based on color characteristic according to claim 1; It is characterized in that: in step (7), introduced Edge Distance and come effectively to reduce wrong merging; Simultaneously some restrictions are carried out in the merging in zone, being defined as of Edge Distance:

P wherein _IjNumber of pixels on the expression edges of regions, x _mAnd x _nRepresent the m of both sides of edges and the position coordinate value that n is ordered respectively, ‖ ‖ representes Euclidean distance.

Images