CN108647692A - Ocean stratification extractive technique based on LH histograms - Google Patents

Abstract

The invention relates to an ocean stratification extraction technology based on temperature and salinity data. Studies have found that the ocean has stratification (ocean stratification refers to the hierarchical structure of seawater density, temperature, salinity and other thermodynamic state parameters distributed with depth). Therefore, we use technologies such as LH histogram and Region Grow to effectively extract ocean stratification in the process of ocean direct volume rendering. The LH histogram is a type of two-dimensional histogram, and its horizontal and vertical axes represent L (lower) and H (higher) values respectively. This technology first processes the ocean temperature and salinity data, calculates the Lower and Higher values of each voxel, and generates a two-dimensional LH histogram; then, selects the ocean stratification on the generated LH histogram; finally, according to the selected results Perform direct volume rendering. If the generated LH histogram does not effectively distinguish the stratification, then a seed point can be selected on the generated histogram, and the Region Grow technology is used to cluster the selected voxels under the given criteria, and the processing After the completion of the direct volume rendering.

Description

Ocean stratification extraction technology based on LH histogram

technical field

The invention belongs to the technical field of ocean information, and in particular relates to an ocean stratification extraction technology based on an LH histogram.

Background technique

Since the discovery of ocean stratification in the 19th century, it has been an important content of physical oceanography research. Oceanic stratification can be divided into thermocline, denocline, halocline, sonic layer, etc. The research results have high academic and military application value. For example, the pyrocline has a great influence on military activities such as submarine activities and anti-submarine warfare. When there is a large positive density gradient in the ocean, that is, "liquid seabed", the submarine can stay on the interface of the layer and keep silent standby. Can effectively evade sonar detection search. Contrary to the "liquid seabed", if the density of the upper layer of seawater is high and the density of the lower layer of seawater is low, it is a density reverse climatic layer, and a "sea cliff" will be formed. When a submarine encounters a "sea cliff", because the buoyancy force on the submarine suddenly decreases, if measures are not taken in time to reduce the weight of the submarine, the submarine will suddenly sink rapidly, causing serious accidents. Because of this, the navies of various countries attach great importance to the research of pyrocline.

Therefore, as an intuitive and effective means and method of data research, data visualization is of great significance to oceanographers studying ocean phenomena. In the early days, people used ocean data to draw temperature-depth line maps and dense-depth line maps to study ocean stratification. Later, with the rapid development of visualization technology, volume rendering technology was applied in the field of oceanography.

There are many kinds of volume rendering techniques. Currently, common data volume rendering methods include ray casting, snowball throwing, and hardware-based 3D texture mapping.

(1) Ray-casting method: Ray-casting method is a volume rendering method that takes the image space as the sequence. It starts from each pixel in the image space and emits a ray in the direction of the line of sight. This ray passes through the three-dimensional data. field, select K equidistant sampling points along this ray, and perform cubic linear interpolation from the color values and opacity values of the 8 data points closest to a certain sampling point to find the opacity value of the sampling point and color values. Then, the color value and opacity value of each sampling point on each ray are synthesized from front to back or from back to front to obtain the color value at the pixel point where the ray is emitted, so that the final image can be obtained on the screen. image of .

(2) Snowball throwing method: Different from the ray-casting method, the snowball throwing algorithm repeatedly performs operations on voxels. It uses a function called Footprint to calculate the influence range of each voxel projection, and uses a Gaussian function to define the intensity distribution (high intensity in the center and small intensity around the periphery), so as to calculate its overall contribution to the image and synthesize it , forming the final image.

(3) Hardware-based 3D texture mapping method: The speed of rendering has always been the main problem of volume rendering, because for the generated image, the processing of each area of the object will affect its quality, so it is necessary to calculate Each voxel, and after voxelization, repositioning the voxel requires a considerable amount of calculation. In order to solve the problem of volume rendering speed, many improved algorithms have been proposed. The common point of these algorithms is to improve the rendering efficiency on the basis of software, such as the algorithm introduced above. The 3D texture mapping method is based on hardware to improve the speed of volume rendering.

At present, there are few studies on the visualization of ocean stratification using volume rendering technology, and most of them use the joint histogram composed of attribute values and attribute gradient values to display the data characteristics. However, in the two-dimensional joint histogram, the boundaries of different substances There will be overlap between them, and the boundary cannot be effectively segmented and extracted.

Contents of the invention

The technical effects of the present invention can overcome the above-mentioned defects, and provide a technique for extracting ocean stratification based on LH histogram. The LH histogram is a type of two-dimensional histogram, and its horizontal and vertical axes correspond to the L (lower) and H (higher) values respectively; the L and H values represent the lower and higher attribute values that constitute the material boundary. In the process of direct volume rendering of the ocean by using the two-dimensional joint histogram, the stratification information of the ocean is not obvious enough in the joint histogram and the boundaries between different substances overlap. We will use the LH histogram to effectively segment the ocean stratification to avoid overlapping of different material boundaries. And the Region Gorw technology can be used to cluster and draw the ocean stratification when the effect of the LH histogram is not ideal.

In order to achieve the above object, the present invention adopts the following technical solutions, and the concrete steps are (taking the profile data of Argo as an example):

(1) Given an epsilon (epsilon is a number greater than 0) threshold;

(2) Calculate the temperature gradient amplitude grad of two adjacent points along the depth direction starting from the first data point of the first profile;

(3) Determine the relationship between the grad of the calculated i-th (i starts from 1) data point and the given threshold, and set the FL and FH values of the data point according to the comparison results;

(4) Continue to calculate the temperature gradient of the i+1th point and perform the comparison and assignment operations in step 3 until a certain condition is met, then stop, and a profile is processed;

(5) Repeat step (4) until all profiles are processed, then stop;

(6) Count the FH and FL of all data points and draw the LH histogram;

(7) Based on the drawn LH histogram, use Region Grow technology to cluster all voxels;

(8) During the rendering process, the opacity of the voxels is set using the gradient size of each data point, so that the voxels closest to the boundary (with the largest gradient) are strengthened.

The beneficial effect of the present invention is that: the LH histogram can avoid the advantages of overlapping different material boundaries to effectively segment and extract the oceanic stratification, and separately visualize the hierarchical structures such as the mixed layer and the cline layer of the ocean, providing more For a rich and intuitive visual experience. In addition, combined with Region Grow technology, different substances can be clustered when the effect of the LH histogram is not ideal, so as to effectively divide the boundaries of substances.

Description of drawings

Fig. 1 Flowchart of marine stratification extraction technology based on LH histogram.

Detailed ways

The marine stratification extraction technology (as shown in Figure 1) based on the LH histogram of the present invention comprises the following steps (taking the profile data of Argo as an example):

(1) Given an epsilon (epsilon is a number greater than 0) threshold;

(2) Calculate the temperature gradient amplitude grad of two adjacent points along the depth direction from the first data point of the first profile, and use the gradient amplitude as the temperature gradient of the two depths corresponding to the two temperatures the gradient value of the mean depth;

(3) Determine the relationship between the grad of the calculated i-th (i starts from 1) data point and the given threshold:

①When grad <= epsilon, it means that the point is located in a homogeneous substance, let FL(i) (Lower value) = FH(i) (Higher value) = the intensity value of this data point, and record the FH of this point ;

②When grad > epsilon, it means that the point is on the boundary, so that the FH(i) of the point is equal to the FH value recorded at the i-1th point;

(4) Continue to calculate the temperature gradient of the i+1th point and perform the comparison and assignment operations in step 3 until it occurs again when the grad of the jth point <= epsilon, let FH(j)=FL( j) = intensity value, and let FL = FL(j) of all points where FL is not assigned a value calculated above;

(5) Repeat step (4) until all profiles are processed, then stop;

(6) According to the FH and FL values of all data points, the points with the same FH and FL values are accumulated and counted;

(7) Draw a two-dimensional LH histogram according to the statistical results of (6);

(8) Select a point on the material boundary of the drawn LH histogram as the seed point of Region Grow, select a certain range in the LH histogram, and then calculate the growth from the seed point to the selected point according to the given rules The required cost value;

(9) According to the cost value calculated in (8), it is judged whether the selected point is located on the boundary where the seed point is located;

(10) During the rendering process, the opacity of the voxels is set using the gradient size of each data point, so that the voxels closest to the boundary (with the largest gradient) are strengthened.

Claims (9)

1. the Ocean stratification extracting method based on LH histograms, specifically includes following basic step：

(1) epsilon is given（Epsilon is the number more than 0）Threshold value, it is characterised in that：Artificially give an epsilon Threshold value, as the fiducial value for judging Ocean stratification；

(2) adjacent along depth direction calculating since first data point of first section by taking Argo cross-sectional datas as an example 2 points of temperature gradient amplitude grad, it is characterised in that：Along between two data points of vertical depth direction calculating of section Temperature gradient amplitude grad, and using the gradient magnitude as the ladder of the mean depth of two depth corresponding with this two temperature Angle value；

(3) judge calculate i-th（I is since 1）The grad of a data point with（1）The magnitude relationship of middle given threshold value, and according to The FL of the data point, FH values is arranged in result of the comparison, it is characterised in that：By the gradient magnitude result of calculating with（1）Given Epsilon values are compared, and then think that the data point is located at the extra large layer of thermohaline uniform properties when gradient magnitude is less than epsilon In, otherwise it is located among the larger spring layer of thermohaline change of properties；

(4) continue to calculate temperature gradient and the execution the of i+1 point（3）The comparison of step and assignment operation, until a section It is disposed, it is characterised in that：According to（3）All data points of step one section of processing, and generate corresponding LH data values；

(5) is repeated（4）Step, until being disposed, all sections then stop, it is characterised in that：Repeat different sections On all data points calculating, compare and amplitude operation；

(6) FH of all data points, FL are counted and draw LH histograms, it is characterised in that：To（5）Walk the LH generated Data are counted, and the number for the data point for including in each pixel in LH coordinate systems is counted, and are then utilized in statistical data Maximum value carry out the normalization of all data, LH grey level histograms are finally drawn according to the result after normalization；

(7) Region Grow technologies are utilized on the basis of the LH histograms drawn, and all voxels are clustered, it is special Sign is：In the case where LH histograms cannot well distinguish Ocean stratification, Region is used in LH spatial domains Grow technologies carry out cluster operation；

(8) in render process, the opacity of voxel is set using the gradient magnitude of each data point, is made near proximal border Voxel（Gradient is maximum）It is strengthened, it is characterised in that：During last data render, provicial commander's size of data point is utilized Different lightness are set, and the opacity that the larger data point of gradient is arranged is larger.

2. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （1）In, threshold value epsilon is the number more than 0, and under normal circumstances, when Hai Shen is less than 200 meters, epsilon values are 0.2, when Value is 0.05 when Hai Shen is more than 200 meters.

3. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （2）In, the gradient magnitude computational methods of data point are that the temperature value of lower data point subtracts the temperature value of upper layer data point, gained Difference divided by lower data point depth value and upper layer data point depth value difference, the result of last gained takes absolute value It is just the gradient magnitude of the data point.

4. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （3）In, judge calculate i-th（I is since 1）The grad of a data point with（1）The magnitude relationship of middle given threshold value：

1. working as grad<When=epsilon, then it represents that the point is located in homogeneous substance, enables FL (i)（Lower values） = FH(i) （Higher values）The intensity value of=the data point, and record the FH of the point；

2. working as grad>When epsilon, then it represents that the point is located on boundary, and the FH (i) of the point is enabled to be remembered equal to (i-1)-th point The FH values of record.

5. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （4）In, continue the temperature gradient for calculating i+1 point and execution the（3）The comparison of step and assignment operation, until working as again J-th point of grad<=epsilon, enables FH (j)=FL (j)=intensity value of the point, and enables all FL calculated above FL=FL (j) of unassignable point.

6. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （5）In, it repeats the calculating of all data points on different sections, compare and amplitude operation.

7. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （6）In, horizontal axis is L values in LH histograms, and vertical reference axis is H values.

8. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （7）In, the seed point a little as Region Grow is selected on the material boundary for the LH histograms drawn, in LH histograms Figure center selects certain range, then calculates the cost risen to from seed point needed for selected point according to given rule Value.

9. the Ocean stratification extracting method according to claim 1 based on LH histograms, which is characterized in that the step （8）In, in last render process, using each data point gradient magnitude be arranged voxel opacity, make near The voxel on boundary（Gradient is maximum）Different lightness it is larger, the boundary of last result between different layers knot is strengthened.