KR100952749B1 - Method for saliency-based lighting for feature emphasis - Google Patents

Abstract

The present invention relates to a saliency-based lighting method that can be implemented by highlighting a feature portion of a 3D image.

According to an exemplary embodiment of the present invention, a illumination based illumination method includes: calculating illumination for each vertex of the three-dimensional image in illuminating the three-dimensional image based on the three-dimensional image; Classifying vertices having more than a predetermined criterion among the calculated salience for each vertex into at least one group; Creating a spotlight region through the classified group; Calculating Gaussian filters for vertices inside the spotlight area and vertices outside the spotlight area; And calculating spotlight area lighting using the Gaussian filter.

Description

Quality-based lighting method to highlight features {METHOD FOR SALIENCY-BASED LIGHTING FOR FEATURE EMPHASIS}

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a illumination based illumination method, and more particularly, to a marine based illumination method that can be embodied by highlighting a feature of a 3D image.

Recently, with the development of technology for producing three-dimensional images, it is possible to generate a three-dimensional image composed of a massive mesh consisting of millions or billions of points. The three-dimensional image is illuminated with a visual realism.

Conventional lighting methods include a local lighting method for calculating only a portion of light directly shining, and a global lighting method for calculating even a portion of light not directly shining. In general, the local illumination method is commonly used in the implementation of three-dimensional images due to advantages in rendering time.

However, the conventional area lighting method has a disadvantage in that feature areas of the 3D image are not highlighted at all because the area of light irradiated from a light source at a specific location is collectively calculated. .

1 is a view showing a three-dimensional molecular image in the form of a sphere using general area illumination.

Referring to FIG. 1, it can be seen that no part can visually identify which part is a characteristic part of the three-dimensional molecular image.

Accordingly, an object of the present invention is to provide a saliency-based lighting method that can be implemented by highlighting a feature portion of a 3D image.

In order to achieve the above object, according to an embodiment of the present invention, in the illumination-based illumination method of illuminating a three-dimensional image based on the saliency, the saliency for each vertex of the three-dimensional image is calculated. Making; Classifying vertices having more than a predetermined criterion among the calculated salience for each vertex into at least one group; Creating a spotlight region through the classified group; Calculating Gaussian filters for vertices inside the spotlight area and vertices outside the spotlight area; And calculating a spotlight area lighting value using the Gaussian filter.

Vertices with more than the predetermined criterion are classified into each group through the K-means clustering method.

Generating the spotlight region includes obtaining a central axis of the classified group through a principal component analysis method, and forming a planar shape using the two largest axes of the central axis, wherein the spotlight region is The flat form.

The planar shape includes an elliptic shape.

The saliency-based illumination method includes calculating a general illumination value for the 3D image; And adding the calculated spotlight area illumination value to the general illumination value.

Computing the salience for each vertex of the three-dimensional image comprises the steps of: generating a low resolution mesh that simplifies the mesh constituting the three-dimensional image; And calculating the sailivity for each vertex of the low resolution mesh.

Each vertex of the three-dimensional image is a vertex of the mesh constituting the three-dimensional image.

Each vertex of the three-dimensional image is a vertex of the low resolution mesh.

According to the saliency-based illumination method of the present invention, it is possible to detect a characteristic part of the three-dimensional image, and apply spotlight illumination different from the surroundings to the detected feature part.

In addition, according to the present invention, by applying a Gaussian filter between the characteristic part and its surroundings, it is effective to realize more natural lighting in the three-dimensional image by using the spotlight illumination gradually toward the characteristic part.

Other objects and advantages of the present invention in addition to the above object will be apparent from the description of the preferred embodiment of the present invention with reference to the accompanying drawings.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to FIGS. 2 to 8.

Prior to this, terms or words used in the specification and claims should not be construed as having a conventional or dictionary meaning, and the inventors should properly explain the concept of terms in order to best explain their own invention. Based on the principle that can be defined, it should be interpreted as meaning and concept corresponding to the technical idea of the present invention. Therefore, the embodiments described in this specification and the configurations shown in the drawings are merely the most preferred embodiments of the present invention and do not represent all the technical ideas of the present invention. Therefore, It is to be understood that equivalents and modifications are possible.

2 is a flowchart illustrating a warranty-based lighting method according to an embodiment of the present invention.

2, the lighting method according to an embodiment of the present invention first goes through the process of obtaining a general lighting value for the three-dimensional image (S1). Here, the general illumination value means an illumination value uniformly applied to the 3D image without highlighting the feature according to the conventional method, and thus, the detailed description thereof will be omitted.

Next, in order to detect the feature point of the 3D image, the salience is calculated (S2). In this case, the salience refers to a difference value between neighboring vertices of a specific vertex of the 3D image.

In the present invention, the center-surround mechanism is applied to a normal vector in order to obtain the saliency.

First, in order to obtain the difference between normal distributions, as shown in FIGS. 3A and 3B, the planes corresponding to the two points are obtained using the Principal Component Analysis (PCA) method, respectively. Calculate the angular difference between the normal vectors of each of the two planes as shown in c). Then, through Equation 1 using the angle difference between the normal vectors, the saliency for a specific vertex at a specific scale is calculated.

Where S _i denotes the seriancy of vertex v at scale i, M is defined as an operation that maps each vertex of the mesh to a normal vector of vertices, and V (M (v), σ _i ) Mean a normal vector of plane containing the points σ _i from the vertex v. On the other hand, σ _i means the distance to the surrounding points in the scale i, the vertex means each vertex of the mesh constituting the three-dimensional image.

That is, the salience for a particular vertex at a particular scale is calculated as the absolute value of the normal vector angle difference.

Then, to calculate the final three days Varian upon by obtaining the three days during Varian for each σ _i σ _i while changing the value. For example, the value of σ _i can be changed in units of 2 ms, 4 ms, and 8 ms, and the final saliency S is calculated through Equation 2 below.

Meanwhile, in the present invention, a method of changing the resolution of a 3D image may be used to improve the computational speed of the sailivity.

Table 1 below is a table showing the time required to calculate the saliency in seconds when the saliency of the 3D image having the vertices of 13K, 26K, and 51K is measured at 2 s, 4 s, and 8 s, respectively.

 2Å  4Å  8Å  13K  2.00  10.34  49.57  26K  6.54  37.23  206.06  51 K  24.51  146.61  929.71

Referring to Table 1, it can be seen that in the case of 13K in the 8 kHz scale, the time of 49.67 seconds, while in the 51K it takes 929.71 seconds, about 19 times longer than 13K.

In spite of the difference in time required, referring to FIG. 4, which shows the results of the calculation of the salience in 13K, 26K, and 51K, respectively, it can be seen that there is almost no difference visually.

Using this result, the present invention generates a simplified low-resolution mesh by lowering the resolution of the mesh constituting the original three-dimensional image to calculate the saliency for it, and calculate the saliency of the original three-dimensional image. By using a method of controlling the lighting, it is possible to shorten the time required for lighting of the 3D image.

Through this process, when the saliency is calculated, it is possible to detect the point where the saliency is high. Since the higher the saliency of a particular vertex, the particular vertex is characteristic of the vertex than the surrounding vertex. At this time, the criterion for high and low of the sailency is based on the set threshold.

Thereafter, the points with the higher the salary are classified, so that a spotlight group of lighting as shown in FIG. 5 is generated (S3). In FIG. 5, each of the points represents a point of high salivity, that is, feature points of the 3D image.

The K-Means Clustering method may be used as a method for generating spotlight groups of illumination. The K-means clustering method is a method of clustering several data into K groups, and since the method is widely used, a detailed description thereof will be omitted.

Through each of the generated spotlight group, as shown in Figure 6, a spotlight area is generated (S4). In the present invention, in the illumination of the three-dimensional image, in order to achieve a natural spotlight effect, spotlight points are formed in an ellipse shape.

For this purpose, the central axis of each spotlight group is obtained through the PCA method. This central axis consists of three axes and forms an ellipse using the two largest axes of the three axes.

After the spotlight region is generated, a Gaussian filter is calculated (S5). The Gaussian filter is a filter for characteristically configuring the lighting of the spotlight region and allowing the average illumination, that is, the general illumination value, to be gradually applied toward the periphery of the spotlight region, and can be calculated by the following equation.

The Gaussian filter of the internal spotlight area vertices have the value 1, of the outer vertices are to be designed more a distance far from the spot region so as to have a low value, the spot area m for any vertex v _j Gaussian filter (G _m ).

Here, d ₁ is any vertex v _j and when that the coordinate average of the vertices in the spotlight area m c _m, c _m and v _j the mean distance of the first axis direction, and the d ₂ is c _m and v _j Means the distance in the second axis direction, k ₁ and k ₂ means the size of the first axis and the second axis, respectively.

Thus, a Gaussian filter as shown in FIG. 7 is calculated centering on the high spotlight area.

After the Gaussian filter is calculated, the spotlight area lighting value is calculated (S6).

For this purpose, the direction of illumination is first set. At this time, the direction of illumination is set to the normal vector average of the vertices located in each spotlight area. Then, the spotlight area lighting value S _j at each vertex v _j is calculated by the following equation.

Here, n _j denotes a normal vector of vertex v _j , L _m denotes an illumination direction of the spotlight region m, and α denotes a weight of illumination. At this time, the weight is a value defining the degree of application of the illumination.

Through this, after the spotlight area light value is calculated, the calculated spotlight area light value is added to the general light value for the initially calculated three-dimensional image and is applied (S7). Is completed.

The spotlight area lighting value according to the present invention may be configured to be darker than the lighting of the surrounding area or may be configured to be brighter than the surrounding area as shown in FIG. 8 according to the setting. Referring to FIG. 8, it can be seen that the characteristic part of the three-dimensional molecular image appears reliably compared to FIG. 1.

Those skilled in the art will appreciate that various changes and modifications can be made without departing from the technical spirit of the present invention. Therefore, the technical scope of the present invention should not be limited to the contents described in the detailed description of the specification but should be defined by the claims.

1 is a view showing a three-dimensional molecular image in the form of a sphere using general area illumination,

2 is a flowchart illustrating a saliency-based lighting method according to an embodiment of the present invention;

3 is a diagram illustrating a process of calculating an angle difference of a normal vector using a principal component analysis method;

4 is a diagram illustrating a result of saliency calculation for each resolution;

5 shows a spotlight group;

6 shows a spotlight area,

7 illustrates a Gaussian filter according to an embodiment of the present invention, and

8 is a view showing a three-dimensional molecular image implemented in accordance with an embodiment of the present invention.

Claims (13)

In illuminating three-dimensional images based on saliency, Calculating a salience for each vertex of the three-dimensional image; Classifying vertices having a saliency greater than or equal to a predetermined threshold value among the calculated saliency for each vertex into at least one group; Creating a spotlight region through the classified group; Calculating Gaussian filters for vertices inside the spotlight area and vertices outside the spotlight area; Calculating a spotlight area lighting value using the Gaussian filter; Calculating a general illumination value for the three-dimensional image; And Adding the calculated spotlight area illumination value to the general illumination value Marine based lighting method comprising a. delete delete The method of claim 1, Said vertices with saliency above the predetermined threshold are classified into each group through a K-means clustering method. The method of claim 4, wherein Generating the spotlight area, Obtaining a central axis of the classified group through a principal component analysis method, and Forming a planar shape using two axes among the central axes Including, Said spotlight area is said planar shape. The method of claim 5, Said planar shape is a marine based illumination method characterized in that it comprises an ellipse form. The method according to claim 5 or 6, Said gaussian filter is calculated according to the following formula.

Here, G _m represents a Gaussian filter of the spotlight area m for any vertex v _j is, d ₁ is to say a coordinate average of the vertices c _m at any vertex v _j and spot area m, c _m and v _j means the distance in the first axis direction, d ₂ means the distance in the second axis direction of c _m and v _j , k ₁ and k ₂ means the size of the first axis and the second axis, respectively. The method of claim 7, wherein Said spotlight area lighting value is calculated according to the following formula.

Where n _j denotes the normal vector of vertex v _j , L _m denotes the lighting direction of the spotlight area m, and α denotes the weight of the illumination. The method of claim 8, The illumination direction of the spotlight zone m is set to a normal vector average of vertices located within the spotlight zone m. delete The method of claim 1, Computing the salience for each vertex of the three-dimensional image, Generating a low resolution mesh that simplifies the mesh constituting the 3D image; And Calculating a salary for each vertex of the low resolution mesh Marine based lighting method comprising a. The method of claim 1, Each vertex of the three-dimensional image is a marine-based lighting method, characterized in that each vertex of the mesh constituting the three-dimensional image. The method of claim 11, Said vertices of said three-dimensional image are each vertex of said low resolution mesh.

Images