CN104778678A - Pore throat recognition method with consideration of pore throat tail end - Google Patents

Abstract

The invention relates to a pore throat recognition method with consideration of a pore throat tail end and in particular to a method for recognizing pores and throats in microscopic visualization model oil displacement experiment images in the field of oilfield development. The method comprises the following steps: carrying out binaryzation treatment on images; carrying out length transformation on space of pores and throats; extracting a center line by using a turning point algorithm; cancelling inner paths in the pores to obtain path curves of the rest of paths; carrying out wavelet noise reduction on the path curves; finding bottleneck positions of the pores and the throats and carrying out pore throat recognition without consideration of the tail end; extracting a center line by using a combustion algorithm and finding the pore throat tail end; extracting a center line of the tail end by using the turning point algorithm by means of an endpoint of a combustion center line of the tail end; and combining the center line of the turning point algorithm and carrying out the pore throat recognition method with consideration of the pore throat tail end. With combination of the combustion algorithm and the turning point algorithm, the tail end part can be independently recognized; the extracted center line is good in centering effect; the relative errors of recognition of the pore throat length are reduced.

Description

A kind of pore constriction recognition methods considering pore throat end

Technical field

The present invention relates to a kind of pore constriction recognition methods considering pore throat end, the particularly hole of microcosmic Visualization Model oil displacement experiment image and the recognition methods of venturi in a kind of oil-field development field.

Background technology

Hole and venturi are the major reservoir spaces of fluid in oil-field development, and position is deposited in the tax of research remaining oil in pore throat and oil and water zonation form is significant to improving oil recovery factor further.Pore throat end refers to the pore throat space only having a passage to be connected with other pore throat in porous medium, and fluid can arrive but not easily circulate, and easily makes crude oil be detained and forms remaining oil.Pore throat end is one of important area of Tapping Residual Oil, identifies that pore throat end is very necessary in microcosmos experiment image.

The venturi knowledge method for distinguishing of current image is mainly the algorithm based on center line, and can its recognition effect depend on to a great extent and accurately extract center line, and it represents algorithm turning point algorithm and burning algorithm.Turning point algorithm is transformed to basis with length, and therefore center line centering effect is better, but automatically cannot extract the center line of pore throat end.Burning algorithm is refined as principle with homotopy, can the topological structure of retaining space, identify end exactly, but the center line centering effect extracted is limited, causes the throat length of identification too short.Therefore in order to reach best pore throat recognition effect, accurate recognition methods both should can identify end, retaining space topological structure, and center line can be made again to reach good centering effect, significantly to reduce the relative error of venturi identification length.

Summary of the invention

For making up the deficiencies in the prior art, to burn algorithm and turning point algorithm of the present invention combines, propose a kind of pore constriction recognition methods that can identify pore throat end, its identifying accurately can identify end and significantly reduce the relative error of the throat length after identifying and physical length.

I. technical solution of the present invention concrete steps are as follows:

(1) binary conversion treatment is carried out to image, distinguish rock particles and pore throat space;

(2) be that foreground pixel carries out European length conversion to pore throat space with rock particles, length transformation calculations formula is:

$D_i=\min \left(D(p(x_i,y_i),q(x_j,y_j))\right)=\min \left(\sqrt{{(x_i-x_j)}^2+{(y_i-y_j)}^2}\right)$

Wherein p be in pore throat space a bit, q is arbitrary rock particles point around p point.

(3) extract image center line by the turning point algorithm converted based on length, center line does not now consider end;

(4) based on turning point center line, pore throat identification during end is not considered;

(5) utilize burning algorithm to extract center line, search burning center line tail end end points, determine end region;

(6) by burning center line tail end end points, turning point algorithm is utilized to extract end center line;

(7) the end-turn point center line in the turning point center line in step (4) and step (6) is merged, pore throat identification is carried out to stub area, then do not consider that the pore throat recognition result of stub area merges by end recognition result and step (4), obtain the pore throat recognition result considering end.

In described step (4), the step of pore throat identification is as follows:

1. delete and be positioned at the turning point median path of pore interior, pixel corresponding to residue median path is numbered in order, and using the length transformation results of each pixel as pore radius corresponding to this pixel;

2. be horizontal ordinate with pixel number, corresponding pore radius is ordinate, make every paths along journey curve;

3. signal will be regarded as along journey curve, discrete Dmey ripple is selected to be wavelet basis function, two Scale Decompositions are carried out to signal, reasonable threshold value is selected to intercept wavelet coefficient, then reconstruction signal complete wavelet noise (list of references: Sun Yankui. wavelet analysis and application [M] .DynoMediaInc. thereof, 2005.), Threshold selection formula is:

wherein σ is noise criteria variance, and N is signal length;

4. after denoising along two local minimums journey curve searched near two ends, using the pixel of its correspondence as hole and the separation of venturi and the bottleneck of venturi, draw pore throat identification line in the position that bottleneck is corresponding;

5. non-medium space be divide into some closed regions by pore throat identification line and rock border, and will comprise the region of center line node as hole, remainder is as venturi.

In described step (6), extract end center line step as follows:

1. from the hole after step (4) identifies or venturi, extract and comprise the venturi of line endpoints pixel in burning or pore region is stub area;

2. with the end pixels of the center line that burns for initial point, be four subgraphs along orthogonal directions by the stub area cutting of extraction, application turning point algorithm extracts the center line of four subgraphs respectively;

3. combined by the center line of four subgraphs and clear up branch, acquired results is the branch of turning point center line in end.

The present invention has following beneficial effect and advantage:

(1) with the center line of turning point algorithm for main algorithm extraction pore space, the overall centering effect of center line is improved;

(2) utilize after wavelet noise along journey profile lookup bottleneck, the throat length after identification and actual throat length relative error significantly reduce;

(3) end points obtained by burning algorithm guides turning point algorithm to extract the center line of end section, end section can be identified separately.

Accompanying drawing explanation

Fig. 1 is flow chart of steps of the present invention.

Fig. 2 is the turning point center line of microcosmic Visualization Model topography.

Fig. 3 is the turning point center line deleting pore interior path, deletes path oval marks.

Fig. 4 a be before wavelet noise certain path along journey curve.

Fig. 4 b be after wavelet noise certain path along journey curve.

Fig. 5 a is pore throat separation and identifies line schematic diagram.

Fig. 5 b is the identified region of pore throat recognition result when not considering end, rectangle marked region corresponding diagram 5a.

Fig. 6 is the burning algorithm center line of Visualization Model topography, line endpoints sphere shaped markup in end.

Fig. 7 a is the end schematic diagram of line endpoints in band burning.

Fig. 7 b is end subgraph identification schematic diagram.

Fig. 7 c is certain the end center line after cleaning branch.

Fig. 8 is the turning point center line after merging with end center line.

Fig. 9 is pore throat recognition result when considering end.

Embodiment

By reference to the accompanying drawings and embodiment the invention will be further described:

As shown in Figure 1, utilize a kind of pore constriction recognition methods that can identify pore throat end, its step is as follows:

(1) binary conversion treatment is carried out to image, distinguish rock particles and pore throat space;

(2) be that foreground pixel carries out European length conversion to pore throat space with rock particles, length transformation calculations formula is:

$D_i=\min \left(D(p(x_i,y_i),q(x_j,y_j))\right)=\min \left(\sqrt{{(x_i-x_j)}^2+{(y_i-y_j)}^2}\right)$

Wherein p be in pore throat space a bit, q is arbitrary rock particles point around p point.

(3) extract image center line by the turning point algorithm converted based on length, Fig. 2 is the turning point center line of certain Visualization Model topography, can find out the unidentified end of center line now;

(4) based on turning point center line, pore throat identification during end is not considered;

1. the length transformation results of each pixel is referred to as pore radius corresponding to pixel for this reason.Calculate the air line distance between two nodes, then calculate radius sum corresponding to two nodes, if air line distance is less than radius sum, then this path is that deletion is should give in pore interior path, as shown in Figure 3.

2. make every paths along journey curve, signal will be regarded as along journey curve, and select discrete Dmey ripple to be wavelet basis function, two Scale Decompositions are carried out to signal, select reasonable threshold value intercept wavelet coefficient, then reconstruction signal completes wavelet noise.Threshold value T selection formula is wherein σ is noise criteria variance, and N is signal length.The general value of threshold value T is 1 ~ 3.5.Fig. 4 a and Fig. 4 b be before and after certain paths denoising along journey curve, wavelet coefficient intercept threshold value be 2.9.

3. as shown in Figure 4 b, after denoising along two local minimums journey curve searched near two ends, using the pixel of its correspondence as hole and the separation of venturi and the bottleneck of venturi, draw pore throat identification line in the position that image bottleneck is corresponding, wherein identify that line is the line of separation from the nearest point of both sides rock.Fig. 5 a is the identification schematic diagram of part hole and venturi, crosses and identifies that the round dot of line represents bottleneck.

4. non-medium space be divide into some closed regions by pore throat identification line and rock border, and will comprise the region of center line node as hole, remainder is as venturi.Fig. 5 b is recognition result when not considering end, and the dash area wherein in pore throat space represents hole, and non-shaded portion represents venturi.

(5) utilize burning algorithm to extract center line, search burning center line tail end end points, determine end region, Fig. 6 is the burning center line of microcosmic Visualization Model topography.

(6) turning point algorithm is utilized to extract end center line;

1. from the hole after step (4) identifies or venturi, extract and comprise the venturi of line endpoints pixel in burning or pore region is stub area, Fig. 7 a be certain end schematic diagram, the some representative of afterbody burn in line endpoints;

2. with the end pixels of the center line that burns for initial point, be four subgraphs along orthogonal directions by the stub area cutting of extraction, need during cutting four images respectively to the border of four width images to external expansion 2 pixels, namely the common edge of four width images has overlapped part separately, application turning point algorithm extracts the center line of four subgraphs respectively, and Fig. 7 b is four subgraphs after certain end identification and center line thereof;

3. combined by the center line of four subgraphs and clear up branch, acquired results is the branch of turning point center line in end, and Fig. 7 c is the center line of certain end after cleaning branch.

(7) the end-turn point center line in the turning point center line in step (4) and step (6) is merged, carry out the pore throat identification considering end, pore throat recognition result in recognition result and step (4) is merged, obtains pore throat recognition result when considering end.Fig. 8 is the turning point center line after end center line and Fig. 3 center line merge, and Fig. 9 is the pore constriction recognition result after end recognition result and Fig. 5 b merge.

Claims (3)

1. consider a pore constriction recognition methods for pore throat end, it is characterized in that following steps:

(1) binary conversion treatment is carried out to image, distinguish rock particles and pore throat space;

(2) be that foreground pixel carries out European length conversion to pore throat space with rock particles, transformation calculations formula is:

$D_i=\min \left(D(p(x_i,y_i),q(x_j,y_j))\right)=\min \left(\sqrt{{(x_i-x_j)}^2+{(y_i-y_j)}^2}\right)$

Wherein p be in pore throat space a bit, q is arbitrary rock particles point around p point;

(3) extract image center line by the turning point algorithm converted based on length, center line does not now consider end;

(4) based on turning point center line, pore throat identification during end is not considered;

(5) utilize burning algorithm to extract center line, search burning center line tail end end points, determine end region;

(6) by burning center line tail end end points, turning point algorithm is utilized to extract end center line;

(7) the end-turn point center line in the turning point center line in step (4) and step (6) is merged, pore throat identification is carried out to stub area, then do not consider that the pore throat recognition result of stub area merges by end recognition result and step (4), obtain the pore throat recognition result considering end.

2. a kind of pore constriction recognition methods that can identify pore throat end as claimed in claim 1, is characterized in that:

In step described in it (4), the step of pore throat identification is as follows:

1. delete and be positioned at the turning point median path of pore interior, pixel corresponding to residue median path is numbered in order, and using the length transformation results of each pixel as pore radius corresponding to this pixel;

2. be horizontal ordinate with pixel number, corresponding pore radius is ordinate, make every paths along journey curve;

3. will regard signal as along journey curve, and select discrete Dmey ripple to be wavelet basis function, carry out two Scale Decompositions to signal, select reasonable threshold value to intercept wavelet coefficient, then reconstruction signal completes wavelet noise, and Threshold selection formula is:

wherein σ is noise criteria variance, and N is signal length;

4. after denoising along two local minimums journey curve searched near two ends, using the pixel of its correspondence as hole and the separation of venturi and the bottleneck of venturi, draw pore throat identification line in the position that bottleneck is corresponding;

5. non-medium space be divide into some closed regions by pore throat identification line and rock border, and will comprise the region of center line node as hole, remainder is as venturi.

3. a kind of pore constriction recognition methods that can identify pore throat end as claimed in claim 1, is characterized in that:

In step described in it (6), extract end center line step as follows:

1. from the hole after step (4) identifies or venturi, extract and comprise the venturi of line endpoints pixel in burning or pore region is stub area;

2. with the end pixels of the center line that burns for initial point, be four subgraphs along orthogonal directions by the stub area cutting of extraction, application turning point algorithm extracts the center line of four subgraphs respectively;

3. combined by the center line of four subgraphs and clear up branch, acquired results is the branch of turning point center line in end.