CN102920436A - Processing method of muscle image by utilizing Hough transform - Google Patents

Abstract

The invention provides a processing method of a muscle image by utilizing Hough transform, which is to automatically mark positions of upper myolemma and lower myolemma by adopting Hough transform, and calculate the thickness between upper myolemma and lower myolemma by design algorithm. The method comprises the following steps: according to the characteristics of a muscle image, at least finding out front two peak points in a Hough matrix so as to detect positions of upper myolemma and lower myolemma in the image; extracting straight lines in positions of the upper myolemma and lower myolemma; according to information of the marked straight lines, calculating the sum of distances of corresponding points in the same row of upper and lower straight lines to acquire the area of muscle; and calculating the average thickness of muscle by dividing the muscle area by the muscle length, wherein the muscle length is the width of the image. According to the image processing method, the thickness of muscle can be quickly and accurately measured, and can be integrated into ultrasonic acquisition equipment for real-time measurement of muscle thickness.

Description

Utilize the muscle image processing method of Hough transformation

Technical field

The present invention relates to a kind of image processing method, relate in particular to a kind of muscle image processing method based on the B ultrasonic collection that utilizes Hough transformation.

Background technology

Muscle is the vital tissue that consists of human body, it be distributed in each histoorgan and skeleton around, its function is shunk and guided-moving for producing, playing the part of at the volley vital effect, and the formation of muscle is very complicated, and quantitative analysis and assessment muscle function state are difficult point and the focuses in sports medical science and the motion function rehabilitation research.

Ultra sonic imaging be a kind of in real time, noinvasive and portable formation method, since emerging, be widely used in each scientific research field.Ultrasonic is the imaging technique that the first can the auxiliary diagnosis muscle disease, along with ultrasonic technique be tending towards ripe, a large amount of researcheres utilizes two-dimensional ultrasonic image diagnosis muscle disease, such as muscle sacred disease, muscle malignant tumor, muscle hematoma and muscle tear etc.From the nineties in last century, there is the scholar to begin to utilize the ultrasonic functional status that removes to assess quantitatively muscle, and analysis result is applied in the research field of biomechanics, such as the functional study of muscle, human motion analysis, the hardness of muscle (elasticity) measurement etc.

Utilize two-dimensional ultrasound can obtain human muscle's image and also analyze the structural parameter that obtains muscle, assess the functional status of muscle with this.Main muscle area of section, the thick wide ratio of cross section, muscle fiber length, muscle thickness and the pinniform angle isostructuralism parameter of adopting explained the change of state of muscle.Wherein, muscle thickness is an important parameter, and therefore how measuring quickly and accurately muscle thickness seems particularly important.The at present measurement for muscle thickness only limits to measure manually, and it lacks objectivity, and certainty of measurement is difficult to control, and for measuring large batch of muscle thickness, operating process is wasted time and energy.

There is report to point out that skeletal muscle thickness can be calculated out by corresponding pinniform angle information.Have document to propose the semi-automatic method that a kind of muscle fiber orientation is estimated, with Radon transform (Radon Transform) realized muscle pinniform angle from motion tracking and calculating.

Summary of the invention

The present invention is based on above-mentioned prior art complicated operation, shortcoming that precision is low, based on the muscle image that B ultrasonic gathers, proposed a kind of simple to operate, precision is high, the fireballing muscle image processing method that utilizes Hough transformation, it is characterized in that, may further comprise the steps:

(1) muscle image pretreatment: use the greyscale transformation function that pretreatment image is carried out greyscale transformation, adjust the contrast of image;

(2) use maximum variance between clusters according to the gamma characteristic of image image to be carried out binary segmentation, the image segmentation that collects is gone out up and down two sarolemmas;

(3) adopt Hough transformation, find out at least two peak points in the Hough matrix, detect up and down two the sarolemma positions in the image, extract the straight lines row labels of going forward side by side two sarolemma positions, straight line information according to labelling, calculate up and down the corresponding point that are in same column in two straight lines apart from sum, obtain the area of muscle; The average thickness of muscle is:

Wherein, muscle length is the width of image.

The muscle image processing method that utilizes Hough transformation of the present invention can be objective, quick, accurate, the thickness of measuring muscle of high duplication.

Description of drawings

Fig. 1 is the flow chart that one embodiment of the invention is utilized the muscle image processing method of Hough transformation;

Fig. 2 is the pretreatment image of the muscle image of one embodiment of the invention muscle image processing method of utilizing Hough transformation;

Fig. 3 be the muscle image of one embodiment of the invention muscle image processing method of utilizing Hough transformation process and labelling after image;

Fig. 4 is point in the image space of one embodiment of the invention muscle image processing method of utilizing Hough transformation and the straight line antithesis sketch map in the parameter space;

Fig. 5 is straight line in the image space of one embodiment of the invention muscle image processing method of utilizing Hough transformation and the some antithesis sketch map in the parameter space;

Fig. 6 is the cumulative array sketch map in the parameter space of one embodiment of the invention muscle image processing method of utilizing Hough transformation;

Fig. 7 is the dotted line antithesis sketch map under the polar equation of one embodiment of the invention muscle image processing method of utilizing Hough transformation.

The specific embodiment

Come the present invention is described in further detail below in conjunction with accompanying drawing and specific embodiment.

As shown in Figure 1, utilize the flow chart of the muscle image processing method of Hough transformation for the present invention, the present invention is based on the muscle image that B ultrasonic gathers, provide a kind of simple to operate, precision is high, the fireballing muscle image processing method that utilizes Hough transformation, may further comprise the steps:

(1) muscle image pretreatment: use the greyscale transformation function that pretreatment image (as shown in Figure 2) is carried out greyscale transformation, adjust the contrast of image;

(2) use maximum variance between clusters according to the gamma characteristic of image image to be carried out binary segmentation, the image segmentation that collects is gone out up and down two sarolemmas;

(3) adopt Hough transformation, find out at least two peak points in the Hough matrix, detect up and down two the sarolemma positions in the image, extract the straight lines row labels (as shown in Figure 3) of going forward side by side two sarolemma positions, straight line information according to labelling, calculate up and down the corresponding point that are in same column in two straight lines apart from sum, obtain the area of muscle; The average thickness of muscle is:

Wherein, muscle length is the width of image.

Hough transformation of the present invention utilizes a little-duality of line, i.e. and the line that in parameter space, intersects of the some correspondence of image space conllinear, conversely, all straight lines that meet at same point in parameter space have the point of conllinear corresponding with it in image space.

In image space X-Y, the point (x, y) of all conllinear can be expressed as with linear equation:

y＝mx+c (1.1)

Wherein m is the slope of straight line, and c is intercept, and same up-to-date style (1.1) can be written as again:

c＝-xm+y (1.2)

Following formula can be regarded as the straight line equation among the parameter space C-M, and the slope of its cathetus is x, and intercept is y.

Comparison expression (1.1) and formula (1.2) can be found out, the straight line in a bit (x, y) the corresponding parameter space in the image space, and the straight line in the image space is to be decided by a bit (m, c) in the parameter space.The basic thought of Hough transformation is to regard above-mentioned two formulas as point in the image space and the common constraints of the point in the parameter space, and defines thus an a pair of mapping from the image space to the parameter space.As shown in Figure 4, for the point in the image space and the straight line antithesis sketch map in the parameter space, embodied the duality relation between this point-line.Figure is positioned at collinear point in the image space shown in 5 (a), that point on the image cathetus is mapped to cluster straight line in the parameter space through formula (1.2) shown in Fig. 5 (b), point on the straight line in the image space is through behind the Hough transformation, straight line in the corresponding parameter space intersects at a point, this point determines, determines that this position in parameter space namely knows the parameter of image cathetus.Hough transformation is transformed into the test problems to putting in the parameter space to the straight-line detection problem in image space, and statistics is finished Detection task by simply adding up in parameter space.

In concrete computational process, the discrete cumulative array that turns to two dimension of parameter space M-C need to be established this array for (m, c), as shown in Figure 6, establish simultaneously [m _Min, m _Max] and [c _Min, c _Max] be respectively the span of slope and intercept.Putting array A during beginning is zero entirely, then for the given marginal point in each space, allows m get all over [m _Min, m _Max] interior all possible value, and the c that calculates correspondence according to formula (1.2).Again according to the value (establish all and round) of m and c to array element A (m, c)=A (m, c)+1.Behind cumulative the end, by the location positioning parameter m of local peaking's point and the value of c among the detection array A.

If the slope of straight line infinitely great (such as the straight line of x=a form), employing formula (1.2) can't be finished detection, in order correctly to identify and detect the straight line of any direction and optional position, can substitute with the straight line polar equation (1.1) formula:

ρ＝xcosθ+ysinθ (1.3)

Shown in Fig. 7 (a), straight line l in the image space, θ are that l crosses the vertical line of initial point and the angle of x axle positive direction, and ρ is that initial point is to the distance of l.At this moment, parameter space just becomes ρ-θ space, any straight line correspondence in the X-Y space point in ρ-θ space, by formula (1.3) as can be known, a bit corresponding sine curve in ρ-θ space in the X-Y space.If one group of point that is positioned on the straight line that parameter ρ and θ determine is arranged, then each point is a corresponding sine curve in the parameter space, all these curves must meet at point (ρ, θ) shown in Fig. 7 (b).

Equally, need parameter space is carried out discretization in the process of calculating, the center point coordinate of each unit is:

$\left\{\begin{array}{cc}{\mathrm{\θ}}_n=(n-\frac{1}{2})\mathrm{\Δ\θ}& (n=\mathrm{1,2},...N_{\mathrm{\θ}})\\ {\mathrm{\ρ}}_n=(n-\frac{1}{2})\mathrm{\Δ\ρ}& (n=\mathrm{1,2},...N_{\mathrm{\ρ}})\end{array}\right.---\left(1.4\right)$

Δ θ=π/N wherein _θ, N _θθ is cut apart hop count for parameter: Δ ρ=π/N _ρ, N _ρThe hop count of cutting apart of parameter ρ, Be in the image point apart from initial point apart from maximum.Concrete computational process gets final product corresponding parameter displacement to above similar.

Utilize the elementary tactics of Hough transformation detection of straight lines in image to be exactly: to remove the possible track of the reference point in the calculating parameter space by the marginal point in the image space, and in an accumulator, the reference point of calculating is counted, select at last peak value.Hough transformation is in fact a kind of voting mechanism, and the discrete point in the parameter space is voted, if the ballot value surpasses a certain threshold value, then thinking has enough picture point to be positioned on the straight line that this reference point determines.The impact that this method is interrupted by noise and straight line appearance is less.

According to the characteristics of ultrasonic muscle image, need find out the first two peak point at least, just can detect up and down two the sarolemma positions in the image.According to the straight line information of labelling, calculate up and down the corresponding point that are in same column in two straight lines apart from sum, this is the area of muscle, can calculate the average thickness of muscle according to following formula.

Two sarolemmas about having significantly in the image of the present invention, image is carried out pretreatment, comprise enhancing, cut apart, then adopt Hough transformation to mark the sarolemma position, realize directly automatically measuring of muscle thickness, the thickness of muscle can be measured objective, quickly and accurately, and the real-time measurement of carrying out muscle thickness in the existing ultrasound acquisition equipment can be integrated into.

Be understandable that, for the person of ordinary skill of the art, can make other various corresponding changes and distortion by technical conceive according to the present invention, and all these change the protection domain that all should belong to claim of the present invention with distortion.

Claims (1)

1. a muscle image processing method that utilizes Hough transformation is characterized in that, may further comprise the steps:

(1) muscle image pretreatment: use the greyscale transformation function that pretreatment image is carried out greyscale transformation, adjust the contrast of image;

(2) use maximum variance between clusters according to the gamma characteristic of image image to be carried out binary segmentation, the image segmentation that collects is gone out up and down two sarolemmas;

(3) adopt Hough transformation, find out at least two peak points in the Hough matrix, detect up and down two the sarolemma positions in the image, extract the straight lines row labels of going forward side by side two sarolemma positions, straight line information according to labelling, calculate up and down the corresponding point that are in same column in two straight lines apart from sum, obtain the area of muscle; The average thickness of muscle is:

Wherein, muscle length is the width of image.

Images