KR19990016895A - Eye area detection device and method - Google Patents

Abstract

According to the present invention, when extracting an eye region from a face image by using a search window having a size considering the eye region, the search exclusion region is registered, and the eye region search speed is controlled by controlling the set position of the search window with reference to the registered information. The present invention relates to an improved eye area detection device and method. To this end, the present invention, the storage unit for storing information from the search excluded region in the predetermined face image; A search area determining unit determining whether a new search area transmitted from the search location setting unit corresponds to the search excluded area by referring to the information stored in the storage unit; An eye region determiner which determines whether or not the new search region is similar to the eye region when the new search region is not the search excluded region; A search position setting unit controlled by a search region determination unit and an eye region determination unit to set a position of a search window on a predetermined face image; And if it is determined that the new search area is not similar to the eye area, the search exclusion area registration unit registers the coordinate information of the search area as a search exclusion area.

Description

Eye area detection device and method

The present invention relates to an apparatus and a method for detecting an eye region from a face image using a search box, and more particularly, to an apparatus and method for detecting an eye region to improve a search speed by controlling a setting position of a search box.

In order to detect an eye region in a predetermined face image, a method of extracting an eye region using a search window having a rectangular region has been proposed. In this method, the eye area is detected while constantly moving the entire area from the upper left side to the lower right side of the face image by a predetermined amount of movement. However, the amount of movement of the search box is usually set to about 2 pixels to 4 pixels, and each search box checks whether the eye area is individually, so many calculations are made, so it takes a long time to detect the corresponding eye area. have.

The present invention has been made to solve the above-described problem, when extracting the eye region from the face image using a search box having a size considering the eye region, register the search excluded region, search by referring to the registered information It is an object of the present invention to provide an apparatus and method for detecting an eye region, which improves an eye region searching speed by controlling a setting position of a window.

According to an aspect of the present invention, there is provided an apparatus for detecting an eye region, the apparatus for detecting an eye region, the apparatus including: a storage unit configured to store search-excluded region information in a predetermined face image; By comparing the coordinate information of the new search area transmitted from the search position setting unit with the search excluded area information stored in the first storage unit, it is determined whether the new search area corresponds to the search excluded area, and the new search area is the search excluded area. A search region determination unit for outputting a search position reset control signal in the case of? As a result of the determination of the search area determination unit, if the new search area is not the search exclusion area, it is determined whether or not the new search area transmitted from the search area determination unit is similar to the eye area, and after the determination, resets the search position for the new search area. An eye region determiner which outputs a signal; When a search position reset control signal is transmitted from a search region determination unit to a setting position of a search window having a size in consideration of an eye region on a predetermined face image, a new search region may be set starting from an end point of a currently set search window. A search position setting unit for moving the position of the search window so as to set the search position to move the search window by a predetermined amount of movement and to provide a new search region when the search position reset control signal is transmitted from the eye region determination unit; And when the new search area is determined to be not similar to the eye area by the eye area determiner and the coordinate information of the new search area is transmitted from the eye area determiner, writing to the storage unit to register the new search area as the search excluded area. And an exclusion zone registration unit.

In order to achieve the above object, the eye region detection method according to the present invention includes a predetermined search having a size considering the eye region when an eye region detection request command is applied from an operator in the eye region detection method for a predetermined face image. Setting an initial search position on a predetermined face image using a window; A first checking step of checking whether or not the search area of the set position is the search excluded area; As a result of performing the first check step, when the search area is the search exclusion area, the first return step moves the predetermined search window to a position where the end point of the currently set search area can be used as a starting point, and returns to the first check step. ; A second check step of checking whether the search area is similar to the eye area if the search area is not the search exclusion area as a result of performing the first check step; A first registration step of registering the search area as the search excluded area when the search area is not similar to the eye area as a result of performing the second check step; A second registration step of registering as an eye area candidate area when the search area is similar to the eye area as a result of performing the second check step; After performing the first registration step and the second registration step, the predetermined search amount is moved to include a second return step of moving to a predetermined search window, and returns to the first check step.

1 is a functional block of an eye region detection apparatus according to the present invention,

2a illustrates an example of a search box setting according to the present invention;

3 is an operation flowchart of an eye area detection method according to the present invention;

* Explanation of symbols for main parts of the drawings

100: processor 101: search position setting unit

102: search area determination unit 103: eye area determination unit

104: search exclusion area register 110: first storage unit

120: second storage unit

The above and other objects and various advantages of the present invention will become more apparent from the preferred embodiments of the invention described below with reference to the accompanying drawings by those skilled in the art.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of an apparatus for detecting an eye region according to the present invention, in which a search position of a search box set by considering a size of an eye region with respect to a predetermined face image is moved by a predetermined amount of movement or jumped by a search box in a moving direction. Search position setting unit 101 for setting to move to a location, the first storage unit 110 that stores the search exclusion area as coordinate information, the search position setting with reference to the information stored in the first storage unit 110 A search area that determines whether the search area of the search position set in the unit 101 corresponds to the search excluded area, and the eye area for the search area that is not determined as the search excluded area by the search area determiner 102. An eye region determining unit 103 for determining whether the region is similar to the second region, a second storage unit 120 for storing the search region determined as an area similar to the eye region by the eye region determination unit 103 as a candidate region of the eye region, Snow Young If from decision 103 is that the search area information is determined to be not similar to the eye region is composed of the first storage unit 110, the exclusion zone registration unit 104 to register the zone in the exclusion area a. In particular, the search position setting unit 101, the search region determination unit 102, the eye region determination unit 103, and the search exclusion region registration unit 104 may be implemented by one processor 100.

2A to 2F are exemplary views of setting a search box according to the present invention when detecting an eye region.

The operation of the eye region detecting apparatus according to the present invention shown in FIG. 1 will be described in detail with reference to FIG. 2.

First, when a predetermined face image is applied and a user command requesting eye area detection is applied, the search position setting unit 101 selects a search position on the face image by using a search window of a predetermined size that is set in consideration of the eye area. Initially, as shown in FIG. 2A, the area (1) of the upper leftmost part of the face image is set as the search position, and when the search position reset control signal is transmitted from the search region determination unit 102 to be described later, The search box is moved to a search position where the end point of the region 1 is a starting point as shown in FIG. 2C. That is, the search position is moved from the end point of the area where the current search box is set to the start point where the new search box is to be set. At this time, the end point of the currently set search window becomes the x-axis direction when moving in the x-axis direction, and becomes the y-axis direction surface when moving in the y-axis direction.

When the search position reset control signal is transmitted from the eye region determiner 103, the search position setting unit 101 moves by dx or dy from the previous search box setting position as illustrated in FIGS. 2B and 2D. Move the search position so that the position is the starting point of a new search box.

In addition, when the currently set area corresponds to the last area 5 in the x-axis direction as shown in FIG. 2D, the search window is moved to a search position starting from the (0, dy) coordinate. After the search window is moved as described above, when the search position reset control signal is transmitted from the search area determination unit 102, the search window is moved in the x-axis direction such that the (x1, dy) coordinate is a starting point as shown in FIG. 2E. Move by x1 (by x width of the previously set search box). However, after the search window is moved to the (0, dy) coordinate, the search position reset control signal is transmitted from the eye region determining unit 103 so that the (dx, dy) coordinate is the starting point (moved by the predetermined movement amount). Move it.

As such, when the search position reset control signal is applied from the search region determination unit 102 to be described later, the search position setting unit 101 sets the end position of the corresponding moving direction of the currently set search window to be the start point of a new search window. Jumps, and when the search position reset control signal is applied from the eye region determiner 103, the search window is moved by a predetermined amount in the corresponding moving direction.

When the coordinate information and the image information of the search area are applied from the search position setting unit 101, the search area determining unit 102 searches for the search area by referring to the information stored in the first storage unit 110. Determine whether it is an excluded area. That is, since the coordinate information of the area corresponding to the search exclusion area is registered in the first storage unit 110, the search area determination unit 102 registers the coordinate information of the applied search area and the first storage unit 110. The coordinate information is compared to check whether all or part of the search area is included in the search exclusion area.

For example, as shown in FIG. 2A, since the information on the search exclusion area is not registered in the first storage unit 110, the eye area determination unit 103 may not use the search area. Send information about However, as shown in FIG. 2B, when information of the search area is moved by (dx, 0) coordinates while the area (1) is registered as the search excluded area, the search area determination unit 102 receives the applied search. Since a part of the area is included in the area (1), it is determined as a search exclusion area and the search position setting control unit 101 sends the search position reset control signal to the search position setting unit 101 without transmitting the information about the search area to the eye area determination unit 103. send.

However, as shown in FIG. 2C, when the search area information for the area (2) is applied, only the coordinate information of the area (1) is registered in the first storage unit 110, so that the currently authorized area (2) is not applied at all. (2) does not determine the area as the search excluded area, and transmits the search area information for the area (2) to the eye area determination unit 103, and the search position reset control signal to the search position setting unit 101. Do not send.

When the eye region determiner 103 receives information on a predetermined search region from the search region determiner 102, the eye region determiner 103 checks whether the transmitted search region is a similar region to the eye region. As a result of the check, if the search area is determined to be similar to the eye area, the search area is determined as a candidate area of the eye area and registered in the second storage unit 120 and the search position is set by the search position setting unit 101. Send the reset control signal.

However, if the search result is not the area similar to the eye area, the coordinate information of the search area is transmitted to the search excluded area registration unit 104. For example, if it is determined that the area (2) shown in FIG. To send). At this time, the eye region decision unit 103 transmits a search position reset control signal to the search position setting unit 101. Accordingly, the eye region determination unit 103 always transmits a search position reset control signal when the eye region determination process for the predetermined search region applied from the search region determination unit 102 is completed.

The search excluded area registration unit 104 registers a new search excluded area by writing coordinate information of the applied search area to the first storage unit 110. After the search exclusion area is registered as described above, the new search area position set by the search position reset control signal of the eye area determination unit 103 is stored in the information stored in the first storage unit 110 by the search area determination unit 102. Most of the time is determined as the search excluded area.

3 is an operation flowchart of an eye area detection method according to the present invention. A detailed operation will be described with reference to FIG. 1. The following operation is performed when the search position setting unit 101, the search region determination unit 102, the eye region determination unit 103, and the search exclusion region registration unit 104 are operated under a system condition provided in one processor 100. Will be described with an example.

When the eye region detection request command is applied from the operator, the processor 100 proceeds to step 301 to set the initial search position as shown in FIG. 2A. In step 302, it is checked whether the set initial search position is the search exclusion area by referring to the information stored in the first storage unit 110 as in the search area determination unit 102 described above.

As a result of the check, in the case of the search exclusion area, the flow proceeds to step 303 and checks whether the location where the current search window is set is the search window area that can be set last in the face image. In this case, the search window area that can be set last becomes an area where the search box at the bottom right of the face image can be set.

As a result of the check in step 303, if it is not the area that can be finally set, the process proceeds to step 304 and the search position is moved so that a search box starting from the end point as described above of the currently set search area can be set. . In step 302, the above-described process is repeated for the search area in which the search window is newly set.

However, as a result of the check in step 303, if the area is finally set, the process proceeds to step 305, and the eye area detection process is performed using the eye area candidate area registered in the second storage unit 120 in the conventional manner. And end the operation.

On the other hand, if the search result of step 302, the search area is not the search exclusion area, the flow proceeds to step 306 to check whether the search area is similar to the eye area by using the previously proposed methods. As a result of the check, if the search area is similar to the eye area, the process proceeds to step 308 through step 307 to register the area as a candidate eye area in the second storage unit 120 and proceeds to step 309.

However, if the search result of step 306 is not similar to the eye area, the process proceeds to step 310 through step 307 to register the area as a search excluded area. That is, after the coordinate information of the corresponding area is written to the first storage unit 110 and registered as the search excluded area, the process proceeds to step 309.

In step 309, the processor 100 checks whether the location where the current search box is set is the last area. As a result of the check, in the case of the last region, the flow proceeds to step 305 and is driven as described above. However, if the result of the check in step 309 is not the last area, the process proceeds to step 311 and moves to the next search position. At this time, the movement is made by a predetermined movement amount. In order to repeat the above process with respect to the new search area obtained by moving the search window, the process returns to step 302.

As described above, the present invention registers an area that is not similar to the eye area in the detection of the eye area using the search box having the size considering the eye size as the search excluded area, and a part of the position where the search box is newly set is registered. If it is included in the excluded search area information, it is not checked whether it is similar to the eye area, and the position of the next search box that is set next is not moved by a predetermined amount, but by moving the end point of the currently set search box as a starting point. As a result, compared to the past, the area to check whether or not the eye area is similar is reduced, thereby improving the eye area retrieval speed in the face image.

Claims (5)

In the eye region detection device for a predetermined face image, A storage unit for storing search excluded region information in the predetermined face image; The coordinate information of the new search area transmitted from the search position setting unit and the search excluded area information stored in the first storage unit are compared to determine whether the new search area corresponds to the search excluded area, A search region determination unit for outputting a search position reset control signal in this case; As a result of the determination of the search area determination unit, if the new search area is not the search exclusion area, it is determined whether the new search area transmitted from the search area determination unit is similar to the eye area, and then a search position for the new search area. An eye region determiner configured to output a reset control signal; When the search position reset control signal is transmitted from the search region determination unit to a setting position of a search window having a size in consideration of the eye region on the predetermined face image, the new search starting from the end point of the currently set search window. Move the position of the search box so that an area can be set, and if the search position reset control signal is transmitted from the eye region determiner, set the search position to move the search window by a predetermined movement amount to provide the new search area. Search position setting unit for; And If the new search area is determined not to be similar to the eye area by the eye area determiner and the coordinate information of the new search area is transmitted from the eye area determiner, the new search area is written to the storage unit to exclude the search. An eye region detection apparatus, comprising: a search exclusion region registration unit for registering as an region, thereby improving an eye region searching speed. The eye region detecting apparatus of claim 1, wherein the search excluded region information stored in the storage unit comprises coordinate information on the predetermined face image. The eye area detecting apparatus of claim 1, wherein the search area determining unit does not output the search position reset control signal when the new search area is not the search exclude area. In the eye region detection method for a predetermined face image, Setting an initial search position on the predetermined face image by using a predetermined search window having a size considering the eye area when an operator requests an eye region detection request from an operator; A first checking step of checking whether the search area of the set position is a search excluded area; As a result of performing the first check step, when the search area is a search exclusion area, the predetermined search window is moved to a position where the end point of the currently set search area can be used as a starting point, and the process returns to the first check step. A first return step; A second check step of checking whether the search area is similar to the eye area when the search area is not the search exclusion area as a result of performing the first check step; A first registration step of registering the search area as a search excluded area when the search area is not similar to an eye area as a result of performing the second check step; A second registration step of registering as the eye area candidate area when the search area is similar to the eye area as a result of performing the second check step; After performing the first registration step and the second registration step, including the second return step of moving the predetermined search window by a predetermined amount of movement, and returns to the first check step to improve the eye area search speed Eye area detection method characterized in. The method of claim 4, wherein the eye region detecting method comprises: a search region which can be set last on the predetermined face image after performing the first check step, the first registration step, and the second registration step. Checking a third step; Detecting the eye region using information registered as the candidate region when the search region is a search region that can be finally set as a result of performing the third check step; And as a result of performing the third check step, if the corresponding search area is not the last search area that can be set, the step of proceeding to the first return step and the second return step, which are corresponding next steps, are further included. Eye area detection method characterized in that.