JP2009266169A - Information processor and method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To more easily and accurately extract a moving object area. <P>SOLUTION: In a moving object area extraction device 11, a whole screen movement calculation part 22 calculates the whole screen movement expressing movement about the whole of a processing screen that is a processing target about each of a plurality of frames constituting a moving image, and a moving object area extraction part 25 decides whether a block is the moving object area or not based on the whole screen movement and activity that is an index representing complicacy of a change of a luminance inside the block obtained by dividing the processing screen. This invention can be applied to an information processor, for example. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an information processing apparatus, method, and program, and more particularly, to an information processing apparatus, method, and program that can extract a moving object region more easily and accurately.

  Conventionally, many extraction methods for extracting a moving object region, which is a moving object (object) region, from an image have been proposed. Specifically, for example, many methods using motion vector information have been proposed as a method for extracting a moving object region.

  Patent Document 1 discloses a technique for extracting a moving object region from the correlation of motion of adjacent blocks. Patent Document 2 discloses a method of referring to past motion vectors and separating a region having a motion vector similar to them from an input image as a moving body region.

In addition, a so-called background difference method, that is, a method for detecting a moving object by generating a background image in which the moving object is not captured and calculating a difference between the background image and the input image has been proposed (for example, And Patent Document 3).
JP 2004-207786 A JP 2002-27480 A JP 2002-157599 A

  However, in the conventional method using the motion vector information including Patent Documents 1 and 2 described above, generally, the extraction result of the moving object region strongly depends on the accuracy of the motion detection, so that the flat portion or the repetition which is likely to fail the motion detection is repeated. When there is a pattern portion or when the deformation of an object is severe, there is a problem that erroneous detection of an animal body is likely to occur.

  Further, in the conventional background subtraction method including Patent Document 3 described above, generally, the processing tends to be complicated and heavy in order to generate the background. In addition, a static background may be used, but when the background itself moves, that is, for example, when panning / tilting, zooming, rotation, or the like of the camera occurs, it is difficult to accurately generate the background. As a result, there is inevitably a problem that erroneous detection of the moving object is likely to occur.

  The present invention has been made in view of such a situation, and makes it possible to detect a moving object more easily and accurately.

  An information processing apparatus according to an aspect of the present invention includes a full-screen motion calculation unit that calculates a full-screen motion representing a motion of the entire processing screen to be processed for each of a plurality of unit images constituting a moving image, Based on the full screen motion calculated by the full screen motion calculation means and an activity that is an index representing the complexity of luminance change in the block obtained by dividing the processing screen, the block is an animal. Determination means for determining whether or not the body region.

  The information processing apparatus according to one aspect of the present invention calculates an evaluation value for a difference from a search destination on the second unit image for each block of the first unit image, and the evaluation value and the An evaluation value table in which a search destination is associated is generated, and a search destination to be detected is detected from the search destinations included in the evaluation value table based on the evaluation value included in the evaluation value table, and is detected. A motion detection unit that uses the search destination as the motion of each block, and the full screen motion calculation unit calculates the full screen motion from the motion of the block detected by the motion detection unit, and the determination The means may use an evaluation value corresponding to the full screen motion vector destination included in the evaluation value table for each block as a difference from the full screen motion vector destination.

  The information processing apparatus according to one aspect of the present invention may further include activity calculation means for calculating the activity.

  An information processing method or program according to one aspect of the present invention calculates a full screen motion representing a motion of an entire processing screen to be processed for each of a plurality of unit images constituting a moving image, and the calculated Determining whether or not the block is a moving object region based on the whole screen movement and an activity that is an index representing the complexity of the luminance change in the block obtained by dividing the processing screen. A program to be executed by a computer that controls an information processing method or information processing apparatus.

  In one aspect of the present invention, for each of a plurality of unit images constituting a moving image, a full screen motion representing a motion of the entire processing screen to be processed is calculated, and the calculated full screen motion; Whether or not the block is a moving object region is determined based on an activity that is an index representing the complexity of the change in luminance in the block obtained by dividing the processing screen.

  According to one aspect of the present invention, for example, an animal body region can be extracted. In particular, for example, a moving body region can be extracted more easily and accurately.

  Embodiments to which the present invention is applied will be described below with reference to the drawings.

  In the present embodiment, a frame is adopted as a moving image processing unit. However, the processing unit is not limited to a frame, and may be a field, for example. Hereinafter, such a moving image processing unit such as a field or a frame is referred to as a unit image. That is, it should be noted that the present invention can be applied to processing on an arbitrary unit image, and the frame described below is merely an example of the unit image.

  FIG. 1 is a diagram showing a functional configuration example of an animal body region extraction device 11 to which the present invention is applied.

  Each frame constituting the moving image is sequentially input to the moving object region extracting device 11 as an input image.

  The moving image region extraction device 11 extracts a moving object region from the input image, and outputs moving object region information, which is information for specifying the moving object region, to the outside. Hereinafter, a series of processes of such a moving image area extracting apparatus 11 is referred to as an extraction process.

  The moving image region extraction apparatus 11 includes a block matching unit 21, a full screen motion calculation unit 22, an evaluation value table holding unit 23, an activity calculation unit 24, and a moving object region extraction unit 25.

  The block matching unit 21 divides the input image into blocks of an arbitrary size in a lattice shape. This block is a unit for extracting the moving object region. That is, the extraction accuracy of the moving object region is determined by the block size of this block. Therefore, the block size may be arbitrary, but it should be noted that it needs to be determined in consideration of the extraction accuracy of the moving object region.

  The block matching unit 21 performs block matching on each divided block and generates an evaluation value table to be described later.

  It is ideal to use the block size after division by the division as it is as the block size used for this block matching. However, the block size used for block matching may be changed to a different block size.

  Ideally, block matching is performed on all blocks simultaneously in parallel. However, depending on the restrictions on the circuit scale, block matching may be performed block by block by serially processing using one matching circuit.

  In block matching, an evaluation value for a difference from a search vector destination on an adjacent frame is calculated for each block. The search destination range (search range) indicated by the search vector destination may be an arbitrary range. An evaluation value D as an example of an evaluation value of a difference from the search vector destination is expressed by Expression (1).

・・・（１）

... (1)

  In Expression (1) (and Expression (2) described later), x and y indicate the x coordinate and y coordinate in the input image, respectively. (Sx, Sy) indicates a search vector. Sx and Sy indicate the x component and the y component of the search vector (Sx, Sy), respectively. BLOCK indicates the block area after the division.

In addition, I _t (x, y) indicates the pixel value of the point (x, y) on the t-th frame at time t. (X + Sx, y + Sy) indicates a search vector destination for the point (x, y). Therefore, I _{t + 1} (x + Sx, y + Sy) is the search vector destination (x + Sx, y + Sy) on the t + 1 frame at time t + 1, one frame after the t frame. The pixel value is shown.

  That is, the evaluation value D (Sx, Sy) is the following value. For the pixel value of the point (x, y) on the t-th frame and the pixel value of the search vector destination (x + Sx, y + Sy) on the t + 1-th frame, all points ( A value obtained by taking the sum of absolute differences with respect to x, y) is an evaluation value D (Sx, Sy).

  Of course, instead of the sum of absolute differences in equation (1), a sum of squares of differences, a sum of differences, or the like can be used.

  The block matching unit 21 generates, for each search vector (Sx, Sy), an evaluation value table in which the evaluation value D (Sx, Sy) of the difference between the search vector (Sx, Sy) and the search vector destination is associated. To do.

  Then, for each block, the block matching unit 21 detects a search vector that gives the minimum evaluation value from search vectors included in the evaluation value table of the block, and represents the detected search vector as the motion of the block. Let it be a motion vector. In addition, since this motion vector represents a local motion compared with the motion of the whole screen mentioned later, it is hereafter called a local motion vector. The block matching unit 21 supplies the local motion vector thus obtained to the full screen motion calculation unit 22.

  The block matching unit 21 supplies the evaluation value table for which the local motion vector has been detected to the evaluation value table holding unit 23.

  The full screen motion calculation unit 22 performs a statistical process for each frame on the local motion vectors of all the blocks supplied from the block matching unit 21, thereby calculating the overall motion of the processing screen to be processed in the input image. A full screen motion vector to be expressed is calculated and supplied to the evaluation value table holding unit 23.

  In addition, the processing method of the statistical process employ | adopted as the whole screen motion calculation part 22 is not specifically limited. For example, a simple method may be considered in which local motion vectors detected for all blocks are histogrammed and the most frequent vector in the histogram is used as a full screen motion vector. Therefore, in this embodiment, it is assumed that this statistical processing is adopted.

  The full screen motion calculation unit 22 calculates a vector obtained as a result of the statistical processing, for example, the above-described mode vector as a full screen motion vector, and supplies it to the evaluation value table holding unit 23.

  The evaluation value table holding unit 23 holds the evaluation value table supplied from the block matching unit 21 for each block.

  Further, when the full-screen motion vector is supplied from the full-screen motion calculation unit 22, the evaluation value table holding unit 23 sequentially sets a predetermined block as a target block to be noted as a processing target. The evaluation value table holding unit 23 refers to the evaluation value table held for the block of interest, and calculates an evaluation value of the full screen motion vector destination. Note that the evaluation value of the full-screen motion vector destination is an evaluation value for the difference from the full-screen motion vector destination in the target block. Therefore, hereinafter, the evaluation value of the full screen motion vector destination is appropriately referred to as a difference from the full screen motion vector destination. Then, the evaluation value table holding unit 23 supplies the difference from the full screen motion vector destination to the moving object region extraction unit 25.

  Hereinafter, the difference from the full-screen motion vector destination is described as Db. The difference Db from the full-screen motion vector destination can be obtained by substituting the full-screen motion vector (Wx, Wy) into equation (1), where (Wx, Wy) is the full-screen motion vector. ) (Ie, Db = D (Wx, Wy)). The full-screen motion vector destination is expressed as (x + Wx, y + Wy).

  The activity calculation unit 24 calculates the activity of each block and supplies it to the moving object region extraction unit 25. Here, the activity is an index representing the complexity of the luminance change in the block. Activity A as an example of activity is represented by Expression (2).

・・・・・（２）

(2)

In Expression (2), I _t (x, y) represents the pixel value of the point (x, y) on the t-th frame at time t. i and j are integer variables of -1 to 1, respectively. Therefore, I _t (x + j, y + i) represents the pixel value of the point (x + j, y + i) on the t-th frame at time t.

  In other words, the activity calculation unit 24 calculates the activity A by calculating the sum of absolute differences from the adjacent eight pixels for each pixel in the block and integrating the sum of absolute differences for all pixels.

  However, the expression of the activity evaluation value is not limited to the expression (2). For example, instead of the difference from the vicinity of the surrounding 8 as in the expression (2), for example, a difference of only the four vicinity of the top, bottom, left and right, or a sum of squares instead of the sum of absolute values, various expressions depending on applications Can be considered. Hereinafter, the case where the activity evaluation value of Expression (2) is used will be described.

  The activity calculation by the activity calculation unit 24 uses pixels developed in an image memory (not shown) at the time of difference calculation by the block matching unit 21, and is further performed in parallel with the block matching by the block matching unit 21 in the architecture. It is preferable. However, it should be noted that it is not always necessary to calculate the activity in parallel with the block matching.

  The moving object region extraction unit 25 determines whether the target block is the background from the difference from the full screen motion vector destination supplied from the evaluation value table holding unit 23 and the activity supplied from the activity calculation unit 24 for the target block. Determine.

  Specifically, for example, the moving object region extraction unit 25 uses the difference Db with respect to the full-screen motion vector destination for the block of interest and the activity A supplied from the activity calculation unit 24 (3) Is evaluated, and it is determined whether or not the block of interest is the background based on whether the calculated value is positive or negative.

    α × A + β × N−Db (3)

  In Expression (3), N indicates the number of pixels in the block. α and β are coefficients of the difference Db from the activity A and the full screen motion vector destination, respectively.

  If the calculated value of the evaluation formula of Expression (3) is negative (<0), it is determined that the block of interest is a moving object region (that is, the block of interest is not the background). On the other hand, if the calculated value of the evaluation expression of Expression (3) is positive (≧ 0), it is determined that the block of interest is the background (that is, the block of interest is not a moving object region).

  The coefficients α and β are parameters for changing the extraction accuracy of the moving object region, and the values can be changed according to the application and application. It is also possible to calculate the coefficients α and β statistically by calculating backward from a moving image whose motion amount is known. That is, for example, for a moving image whose motion amount is known, the difference Db between the activity A and the full screen motion vector destination is obtained, and statistical processing is performed for each of the block determined as the moving object region and the block determined as the background. By doing so, the coefficients α and β can be obtained.

  When extracting a moving object region from a general moving image, for example, α = 0.8 and β = 2 are preferable.

  Thereafter, when the moving object region extraction unit 25 determines whether all blocks are backgrounds, the moving object region extraction unit 25 extracts each of a set of consecutive blocks among the blocks determined as moving object regions as moving object regions. . Then, the moving object region extracting unit 25 generates moving object region information that is information for specifying the extracted moving object region, and outputs the generated moving object region information to the outside.

  Next, an example of the extraction process of the moving object region extraction device 11 will be described with reference to the flowcharts of FIGS.

  Each frame constituting the moving image is sequentially input as an input image to the block matching unit 21 and the activity calculation unit 24 of the moving object region extraction device 11.

  In step S1, the block matching unit 21 divides the input image into blocks of an arbitrary size in a grid pattern.

  Specifically, for example, the block matching unit 21 divides an input image processing screen into blocks of a predetermined size as shown in FIG. 4 to generate 70 × 7 × 10 blocks. Of course, the entire input image or a screen of another size can be used as the processing screen.

  In step S2, the block matching unit 21 sets a predetermined block among the divided blocks as a target block. Thereafter, the process proceeds to step S3.

  In step S3, the block matching unit 21 performs block matching on the block of interest, and generates an evaluation value table in which the search vector and the evaluation value of the difference between the search vector destinations are associated with each search vector. Thereafter, the process proceeds to step S4.

  In step S <b> 4, the block matching unit 21 detects a search vector that gives the minimum evaluation value from among the search vectors included in the evaluation value table as a local motion vector, and supplies it to the full-screen motion calculation unit 22. Thereafter, the process proceeds to step S5.

  In step S <b> 5, the block matching unit 21 supplies the evaluation value table for the block of interest to the evaluation value table holding unit 23. Thereafter, the process proceeds to step S6.

  In step S <b> 6, the evaluation value table holding unit 23 holds the evaluation value table supplied from the block matching unit 21 for each block. Thereafter, the process proceeds to step S7.

  In step S7, the block matching unit 21 determines whether or not the target block is the last block in the input image.

  If the target block is not the last block in the input image, it is determined as NO in step S7, the process returns to step S2, and the subsequent processes are repeated.

  That is, the loop process of steps S2 to S7 is repeatedly executed for each block constituting the input image. If the evaluation value table is held in the process of step S6 for the last block in the input image, it is determined YES in the process of step S7, and the process proceeds to step S8.

  In step S <b> 8, the full screen motion calculation unit 22 calculates a full screen motion vector by performing statistical processing on the local motion vectors of all blocks supplied from the block matching unit 21, and the evaluation value table holding unit 23. To supply. Thereafter, the process proceeds to step S9.

  In step S9, the evaluation value table holding unit 23 sets a predetermined block in the input image as a target block. Thereafter, the process proceeds to step S10.

  In step S10, the evaluation value table holding unit 23 refers to the evaluation value table for the block of interest and obtains a difference from the full screen motion vector destination. Then, the evaluation value table holding unit 23 supplies the obtained difference from the full-screen motion vector destination to the moving object region extraction unit 25. Thereafter, the process proceeds to step S11.

  In step S <b> 11, the activity calculation unit 24 calculates the activity of the block of interest and supplies it to the moving object region extraction unit 25.

  Specifically, for example, the activity calculation unit 24 obtains the sum of absolute differences from the adjacent 8 pixels for each pixel in the target block, and integrates the sum of absolute differences for all the pixels in the target block. To do.

  That is, as shown in FIG. 5, the activity calculation unit 24 sets a predetermined pixel in the target block as the target pixel. In the example of FIG. 5, it is assumed that the block is composed of M × N pixels of vertical M and horizontal N.

  First, the activity calculation unit 24 sets, for example, the pixel in the first row and the first column in the target block as the target pixel. As shown in the dotted circle at the upper left of FIG. 5, the activity calculation unit 24 includes the eight pixels adjacent to the target pixel, that is, the upper left, upper, upper right, left, right, lower left, lower, and lower right pixels of the target pixel. Find the sum of absolute differences from each. Next, the activity calculation unit 24 sets, for example, the pixel on the right side of the target pixel, that is, the pixel in the first row and the second column as a new target pixel, and obtains the sum of absolute differences from the adjacent eight pixels. In the same manner, the pixel immediately adjacent to the pixel of interest (pixel in the first row and j column) (pixel in the first row and j + 1 (3 ≦ j ≦ N) column) is sequentially set as a new pixel of interest. Then, the sum of absolute differences from the adjacent 8 pixels is obtained. Thereafter, when the calculation is completed for the pixels in the first row, the calculation is performed for the pixels in the lower row of the first row, that is, the pixels in the second row. Similarly, when the calculation for the pixels in the i-th (2 ≦ i ≦ M) row is completed, the calculation is performed for the pixels in the lower row of the i-th row, that is, the pixels in the i + 1-th row. In this way, the activity calculation unit 24 calculates the sum of absolute differences with respect to all the pixels in the block of interest with respect to the adjacent 8 pixels, and integrates the sum of absolute differences of all pixels in the block of interest.

  In step S12, the moving object region extraction unit 25 calculates the difference between the target block full-screen motion vector destination supplied from the evaluation value table holding unit 23 and the activity of the target block supplied from the activity calculation unit 24. It is determined whether the block of interest is a moving object region.

  Specifically, the moving object region extraction unit 25 calculates the difference Db from the full-screen motion vector destination of the target block from the evaluation value table holding unit 23 and the activity A of the target block supplied from the activity calculation unit 24. The used evaluation formula shown in Formula (3) is calculated. Then, the moving object region extraction unit 25 determines that the target block is a moving object region when the calculated value of the evaluation formula is negative, and if the calculated value of the evaluation formula is positive, the target block is not the moving object region. judge.

  In step S <b> 13, the evaluation value table holding unit 23 determines whether the target block is the last block in the input image.

  If the target block is not the last block in the input image, it is determined as NO in step S13, the process returns to step S10, and the subsequent processes are repeated.

  That is, the loop process of steps S9 to S13 is repeatedly executed for each block in the input image. If it is determined in step S12 that the last block in the input image is a moving object region, it is determined YES in step S13, and the process proceeds to step S14.

  In step S <b> 14, the moving object region extraction unit 25 extracts each set of consecutive blocks among the blocks determined as moving object regions as moving object regions. Thereafter, the process proceeds to step S15.

  In step S15, the moving object region extracting unit 25 generates moving object region information, which is information for specifying the extracted moving object region, and outputs it to the outside. Then, the extraction process ends.

  As described above, the moving object region extraction device 11 uses a feature amount such as an evaluation value on the evaluation value table obtained at the time of motion detection as one of the large features for determining the moving object region. As a result, in realizing the extraction of the moving object region, the additional elements to be added to the general motion detection circuit can be extremely reduced in terms of circuit scale and processing man-hours.

  Therefore, it can be said that the moving object region extraction device 11 is very compatible with an application that originally assumed motion detection, for example, an application such as tracking, compression, or creation of spatiotemporal resolution.

  As an example of such an application, for example, a surveillance camera system that tracks an object designated by a user is proposed in Japanese Patent Laid-Open No. 2007-272732.

  Specifically, in this surveillance camera system, a motion vector is used as a feature amount, and a full screen motion or the like is obtained by statistical processing and used for tracking.

  In this surveillance camera system, a full-screen motion vector and an evaluation value table are obtained, and a mechanism for outputting the same activity as the activity calculated by the moving object region extraction device 11 is also used.

  Therefore, when the moving body region extraction apparatus 11 is configured using this monitoring camera system, it is possible to obtain the feature quantities such as the full-screen motion vector, the evaluation value on the evaluation value table, and the activity from the monitoring camera system without an additional circuit. it can. From these full-screen motion vectors and the evaluation values on the evaluation value table, the difference from the full-screen motion vector destination and the output result of activity are obtained. Then, by adding an extremely small circuit for performing the calculation of the evaluation expression of the expression (3) using this output result and the comparison between the calculated value and the threshold value (0 in this embodiment), A moving object can be detected.

  Furthermore, the moving object region information obtained by moving object detection is one of the feature quantities that are very effective for tracking by this surveillance camera system.

  In addition, the moving object region extracting apparatus 11 defines an activity as another major feature, and determines the moving object region using the activity as a feature amount. Since the activity calculation can be realized with an extremely small circuit, the moving object region extracting apparatus 11 can realize a simple yet robust moving object region extraction by effectively utilizing the activity.

  Next, with reference to FIGS. 6 and 7, the effect on the extraction result of the moving object region by considering the activity will be specifically described.

  FIG. 6A (and FIG. 7A described later) shows the extraction result of the moving object region by the moving object region extracting device 11 when the activity is considered. FIG. 6B (and FIG. 7B described later) shows the extraction result of the moving object region by the moving object region extracting device 11 when the activity is not considered as a pair comparison.

  In FIG. 6A and FIG. 6B, the area determined as the moving object area is superimposed on the input image P1 as an edged gray area.

  The input image P1 is one frame of a general TV (television) video signal, and is an image in which panning motion exists and there are many flat and difficult motion detection regions such as a skating rink and a fence. Nevertheless, in FIG. 6A, that is, in the extraction result of the moving object region in consideration of the activity, it can be seen that the skater which is the moving object to be determined as the moving object region is clearly extracted. On the other hand, in FIG. 7B, that is, in the extraction result of the moving object region when the activity is not considered, a part other than the part to be determined as the moving object region, for example, “Smart Ones” located in the back of the skater. It can be seen that misjudgment is conspicuous in the details such as the letters and auditorium seats.

  In FIG. 7A and FIG. 7B, the region determined to be a moving object region is superimposed on the input image P2 as an edged gray region.

  The input image P2 is one frame of a moving image different from the TV video signal. The input image P2 is an image in which there are a plurality of objects to be detected as a moving object or there is a strong detail in the background. Nonetheless, in FIG. 7A, that is, the extraction result of the moving object region in consideration of the activity, the moving object region to be determined as the moving object region is slightly above the right with the child located slightly in the left center. It can be seen that the dogs located are extracted without problems. On the other hand, in FIG. 7B, that is, in the extraction result of the moving object region when the activity is not considered, not only the portion that should be determined as the moving object region but also most of the screen is erroneously determined as the moving object region. As a result, the moving object region cannot be determined at all.

  As described above, the moving object region extracting apparatus 11 can realize highly accurate moving object region extraction by using the feature amount used for motion detection and adding an extremely small additional circuit.

  In addition, the moving object region extraction device 11 is less likely to cause problems such as failure of extraction results in flat portions and repeated pattern portions, which are difficult to detect motion vectors, which are common in conventional methods based on motion vector detection and comparison. it can.

  Furthermore, the moving object region extraction apparatus 11 can realize the extraction of the moving object region without any problem even in the case of a moving image with a moving background such as pan / tilt parallel camera movement.

  That is, the moving object region extracting apparatus 11 can solve the problems of the conventional technology and can extract a moving object region with a simple process and architecture.

  The series of processes described above can be executed by hardware or can be executed by software. When a series of processing is executed by software, a program constituting the software executes various functions by installing a computer incorporated in dedicated hardware or various programs. For example, it is installed from a program recording medium in a general-purpose personal computer or the like.

  FIG. 8 is a block diagram illustrating an example of a hardware configuration of a computer that executes the above-described series of processes using a program.

  In the computer, a CPU 41, a ROM (Read Only Memory) 42, and a RAM (Random Access Memory) 43 are connected to each other by a bus 44.

  An input / output interface 45 is further connected to the bus 44. The input / output interface 45 includes an input unit 46 including a keyboard, a mouse, and a microphone, an output unit 47 including a display and a speaker, a storage unit 48 including a hard disk and a nonvolatile memory, and a communication unit 49 including a network interface. A drive 50 for driving a removable medium 51 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is connected.

  In the computer configured as described above, the CPU 41 loads, for example, a program stored in the storage unit 48 to the RAM 43 via the input / output interface 45 and the bus 44 and executes the program. Is performed.

  The program executed by the computer (CPU 41) is, for example, a magnetic disk (including a flexible disk), an optical disk (CD-ROM (Compact Disc-Read Only Memory), DVD (Digital Versatile Disc), etc.), a magneto-optical disk, or a semiconductor. It is recorded on a removable medium 51, which is a package medium composed of a memory or the like, or is provided via a wired or wireless transmission medium such as a local area network, the Internet, or digital satellite broadcasting.

  The program can be installed in the storage unit 48 via the input / output interface 45 by attaching the removable medium 51 to the drive 50. The program can be received by the communication unit 49 via a wired or wireless transmission medium and installed in the storage unit 48. In addition, the program can be installed in the ROM 42 or the storage unit 48 in advance.

  The program executed by the computer may be a program that is processed in time series in the order described in this specification, or in parallel or at a necessary timing such as when a call is made. It may be a program for processing.

  The embodiment of the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the gist of the present invention.

It is a figure which shows the functional structural example of the moving body area | region extraction apparatus. It is a flowchart explaining an example of the extraction process of the moving body area | region extraction apparatus 11. FIG. It is a flowchart explaining an example of the extraction process of the moving body area | region extraction apparatus 11. FIG. It is a figure which shows an example of a block. It is a figure for demonstrating calculation of activity. It is a figure which shows an example of the extraction result of a moving body area | region. It is a figure which shows an example of the extraction result of a moving body area | region. It is a block diagram which shows the structural example of the hardware of the computer to which this invention is applied.

Explanation of symbols

  11 moving object region extracting device, 21 block matching unit, 22 full screen motion calculating unit, 23 evaluation value table holding unit, 24 activity calculating unit, 25 moving object region extracting unit, 41 CPU, 42 ROM, 43 RAM, 44 bus, 45 I / O interface, 46 input section, 47 output section, 48 storage section, 49 communication section, 50 drives, 51 removable media

Claims (8)

A full screen motion calculating means for calculating a full screen motion representing a motion of the entire processing screen to be processed for each of a plurality of unit images constituting the moving image;
Based on the full screen motion calculated by the full screen motion calculation means and an activity that is an index representing the complexity of luminance change in the block obtained by dividing the processing screen, the block is an animal. An information processing apparatus comprising: determination means for determining whether or not the body region. The determination means determines whether the block is a moving object region based on a difference between the block of the first unit image and a full screen motion vector destination based on the full screen motion on the second unit image. The information processing apparatus according to claim 1. The determination means calculates a predetermined evaluation formula using a difference between the activity of the first unit image and the full-screen motion vector destination on the second unit image, and by the calculated value, The information processing apparatus according to claim 2, wherein it is determined whether or not the block is a moving object region. For each block of the first unit image, an evaluation value for a difference from the search destination on the second unit image is calculated, and an evaluation value table in which the evaluation value is associated with the search destination is generated. Then, a search destination to be detected is detected from search destinations included in the evaluation value table based on an evaluation value included in the evaluation value table, and the detected search destination is a motion of each block. A detection means;
The full screen motion calculating means calculates the full screen motion from the motion of the block detected by the motion detecting means,
The information processing apparatus according to claim 2, wherein the determination unit uses an evaluation value corresponding to the full screen motion vector destination included in the evaluation value table for each block as a difference from the full screen motion vector destination. The information processing apparatus according to claim 1, wherein the activity is an integrated value in the block of a difference sum between each pixel in the block and its adjacent pixels. The information processing apparatus according to claim 1, further comprising activity calculation means for calculating the activity. For each of a plurality of unit images constituting a moving image, calculate a full screen motion representing a motion of the entire processing screen to be processed,
Based on the calculated full-screen motion and an activity that is an index representing the complexity of luminance change in the block obtained by dividing the processing screen, it is determined whether or not the block is a moving object region. An information processing method including a step. For each of a plurality of unit images constituting a moving image, calculate a full screen motion representing a motion of the entire processing screen to be processed,
Based on the calculated full-screen motion and an activity that is an index representing the complexity of luminance change in the block obtained by dividing the processing screen, it is determined whether or not the block is a moving object region. A program to be executed by a computer that controls an information processing apparatus including a step.

Images