JP6930011B2 - Information processing equipment, information processing system, and image processing method - Google Patents

Description

The present invention relates to an information processing apparatus that performs processing using captured images and an image processing method performed by the apparatus.

A game is known in which a part of the body such as the user's head is photographed with a video camera, a predetermined area such as eyes, mouth, and hands is extracted, and the area is replaced with another image and displayed on a display. (See, for example, Patent Document 1). There is also known a user interface system that receives mouth and hand movements taken by a video camera as application operation instructions. In this way, technologies that capture the real world, display a virtual world that responds to its movements, and perform some kind of information processing are used in a wide range of fields, from small mobile terminals to leisure facilities, regardless of their scale. ing.

As a technology to identify information related to the state of a real object using captured images, we introduced a stereo camera that captures the same space from different left and right viewpoints, and based on the parallax of the same subject image in the captured stereo image, the subject A stereo imaging method is known to obtain the distance to. In this technique, in addition to using a general color image in which reflection of ambient light is detected, there is also a case in which an image in which light in a specific wavelength band such as infrared rays is irradiated to the subject space and the reflection is detected is used.

On the other hand, in recent years, as a result of various technological developments for improving image quality in video displays such as television broadcasts and distributed videos, in addition to technologies for improving resolution and color gamut, signals with an expanded brightness range are processed. Technology is becoming widespread. For example, HDR (High Dynamic Range) has a brightness tolerance of about 100 times that of conventional SDR (Standard Dynamic Range), so it is possible to more realistically express objects that feel dazzling in the real world. ..

European Patent Application Publication No. 0999518

The brightness range of the captured image changes variously depending on the combination of the conditions of the subject space such as the position, number, color, pattern, and light condition of the subject and the settings on the image pickup device side such as the shooting conditions and image correction parameters. obtain. For this reason, when information about a subject is obtained or a display image is generated using a captured image, the required accuracy cannot be obtained due to such uncertainties, or the performance of the display device is utilized. It is possible that it cannot be cut off.

The present invention has been made in view of these problems, and an object of the present invention is to provide a technique for acquiring information on a real object using a photographed image with stable accuracy. Another object of the present invention is to provide a technique capable of realizing an image expression in a suitable luminance range by using a photographed image.

One aspect of the present invention relates to an information processing device. This information processing device has an image data acquisition unit that sequentially acquires image data of frames of moving images to be captured, pixel values of newly acquired images of the current frame, and images of past frames acquired before that. An image addition unit that generates an added image in which pixel values are added at corresponding positions, and feature points are extracted from the added image to perform predetermined analysis processing, and feature points are also extracted from the image of the current frame. It is characterized by including an image analysis unit that performs the same analysis processing and integrates the results of both, and an output unit that outputs data representing the integrated results.

Here, the "frame image" may be an image of a frame constituting a moving image periodically taken by one camera, or a frame image constituting a moving image taken simultaneously and periodically by a plurality of cameras. It may be an image. Further, the "predetermined analysis process" performed using the added image may be any general process for performing some output using the captured image. For example, acquisition of position and posture, object recognition, motion detection, visual tracking analysis and the like including processing of detecting feature points from captured images can be mentioned.

Another aspect of the present invention relates to an information processing system. This information processing system includes a head-mounted display equipped with an imaging device that captures a moving image in a field of view corresponding to the user's line of sight, and information processing that generates data of a display image to be displayed on the head-mounted display based on the moving image. The information processing device includes an image data acquisition unit that sequentially acquires image data of moving image frames, and a pixel value of a newly acquired image of the current frame of a past frame acquired before that. An image addition unit that generates an added image in which the pixel values of the image are added at the corresponding positions, and a feature point is extracted from the added image to perform a predetermined analysis process, and the feature point is also extracted from the image of the current frame. It is characterized in that it is provided with an image analysis unit that performs the same analysis processing and integrates the results of both, and an output unit that generates and outputs the data of the display image using the integrated result. ..

Yet another aspect of the present invention relates to an image processing method. In this image processing method, the image data of the frame of the moving image to be captured is sequentially acquired and stored in the memory, and the pixel value of the newly acquired image of the current frame is read from the memory and acquired before that. In addition to the step of generating an added image in which the pixel values of the images of the past frame are added at the corresponding positions, and the feature points are extracted from the added image and subjected to a predetermined analysis process, the feature points are also obtained from the image of the current frame. It is characterized in that it includes a step of extracting and performing the same analysis process and integrating the results of both, and a step of outputting data representing the integrated result.

It should be noted that any combination of the above components and the conversion of the expression of the present invention between a method, a device, a system, a computer program, a recording medium on which a computer program is recorded, and the like are also effective as aspects of the present invention. ..

According to the present invention, it is possible to stably obtain suitable results in the acquisition of position information of a real object and the display of an image using a photographed image.

It is a figure which shows the configuration example of the information processing system of Embodiment 1. It is a figure which shows the example of the appearance shape when the display device of Embodiment 1 is a head-mounted display. It is a figure which shows the internal circuit structure of the information processing apparatus in Embodiment 1. FIG. It is a figure which shows the structure of the functional block of the information processing apparatus in Embodiment 1. FIG. It is a figure for demonstrating the technique of irradiating the pattern with infrared rays in Embodiment 1, and acquiring the distance of a subject using the photographed image of the reflected light. It is a figure for demonstrating the action by adding a frame in Embodiment 1. FIG. It is a figure which shows the state of integrating a plurality of depth images in Embodiment 1. FIG. FIG. 5 is a flowchart showing a processing procedure in which the information processing apparatus acquires position information using captured images and outputs data in the first embodiment. It is a figure which shows the structure of the functional block of the information processing apparatus in Embodiment 2. FIG. It is a figure which shows typically how the image addition part in Embodiment 2 adds with the image of the present frame after correcting the image of the past frame. FIG. 5 is a flowchart showing a processing procedure in which the information processing apparatus according to the second embodiment expands the brightness range of the captured image and displays it. It is a figure which shows typically how the addition part in Embodiment 2 adds an image by shifting a pixel area.

Embodiment 1
The present embodiment relates to a technique for acquiring position information of a subject from a captured image. In many cases, such a technique detects feature points from a captured image. However, the appearance of feature points on the image changes in various ways depending on the situation in the real space such as the brightness of the subject space and the actual position of the subject. Especially in a region where the brightness is low, the feature points cannot be detected, and it is conceivable that the position information becomes indefinite or contains a lot of errors.

In the present embodiment, the brightness range is controlled by adding the captured images of a plurality of frames, and the detection accuracy of the feature points is improved. In the following description, the main focus will be on the method of acquiring the position information of the subject using stereo images obtained by capturing the same space from the left and right viewpoints, but in the present embodiment, the process of detecting feature points from the captured images is performed. If included, it can be applied in the same way, and the target is not limited to stereo images. The purpose of detecting the feature points is not limited to the acquisition of the position information of the subject, but may be any of various image analyzes such as face detection, face recognition, object detection, and visual tracking.

FIG. 1 shows a configuration example of the information processing system of the present embodiment. The information processing system 1 includes an image pickup device 12 that captures a real space, an information processing device 10 that performs information processing based on the captured image, and a display device 16 that displays an image output by the information processing device 10. The information processing device 10 may be connectable to a network 18 such as the Internet.

The information processing device 10, the image pickup device 12, the display device 16, and the network 18 may be connected by a wired cable, or may be wirelessly connected by a wireless LAN (Local Area Network) or the like. Any two or all of the image pickup device 12, the information processing device 10, and the display device 16 may be combined and integrally equipped. For example, the information processing system 1 may be realized by a mobile terminal or a head-mounted display equipped with them. In any case, the external shapes of the image pickup device 12, the information processing device 10, and the display device 16 are not limited to those shown in the drawing. Further, if the image display is not required depending on the content of the information processing, the display device 16 may not be provided.

The image pickup apparatus 12 includes a pair of cameras that capture a subject space at a predetermined frame rate from left and right positions having a known interval. A pair of images taken by the image pickup apparatus 12 from the left and right viewpoints, that is, stereo image data are sequentially transmitted to the information processing apparatus 10. By analyzing the stereo image, the information processing device 10 acquires the position information of the subject in the three-dimensional real space including the distance from the imaging surface. A technique for acquiring the position information of a subject from a stereo image is conventionally known.

That is, the corresponding points representing the images of the same subject are obtained from the pair of images, and the distance from the camera to the subject is obtained by the principle of triangulation using the positional deviation as the parallax. From the position of the image on the image plane and the distance, the position coordinates of the subject in the three-dimensional space can be obtained. The information processing device 10 generates, for example, a depth image in which the distance to the subject obtained by analysis is represented as the pixel value of the image on the image plane as position information.

For this purpose, the type of image captured by the image pickup apparatus 12 is not limited. For example, the image pickup apparatus 12 may capture a color image of visible light by a camera having a general image pickup element such as a CMOS (Complementary Metal Oxide Semiconductor) sensor. Alternatively, light in a specific wavelength band such as infrared rays may be detected and an image showing the intensity distribution of the light may be taken.

In this case, the image pickup apparatus 12 may be provided with a mechanism for irradiating the subject space with light in the wavelength band to be detected. A method of irradiating spot-shaped, slit-shaped, or patterned light and photographing the reflected light to obtain a distance to the subject is known as an active stereo method. Compared with the so-called passive stereo method, which obtains a distance from a color stereo image of ambient light, the active stereo method has a characteristic that it is easy to extract corresponding points in an image even for a subject having few feature points.

When invisible light is used to obtain position information, a camera for capturing a general color image may be separately provided in the image pickup apparatus 12 and used for another purpose such as generation of a display image. Hereinafter, the two-dimensional data of the brightness of the light detected by the image pickup apparatus 12 is collectively referred to as an "image" regardless of the wavelength band of the light to be detected. The information processing device 10 acquires the position information of the subject at a predetermined rate as described above using the image data transmitted from the image pickup device 12, and appropriately performs information processing based on the information processing device 10 to generate output data.

Here, the content of the output data is not particularly limited, and may vary depending on the function requested by the user from the system, the content of the started application, and the like. For example, the information processing device 10 may perform some processing on the captured image or advance the electronic game to generate a game screen based on the position information of the subject. Typical examples of such an embodiment include virtual reality (VR) and augmented reality (AR).

The display device 16 includes a display such as a liquid crystal display, plasma, or organic EL that outputs an image, and a speaker that outputs sound, and outputs output data supplied from the information processing device 10 as an image or sound. The display device 16 may be a television receiver, various monitors, a display screen of a mobile terminal, or the like, or may be a head-mounted display that is worn on the user's head and displays an image in front of the user's head.

FIG. 2 shows an example of the appearance shape when the display device 16 is a head-mounted display. In this example, the head-mounted display 100 is composed of an output mechanism unit 102 and a mounting mechanism unit 104. The mounting mechanism unit 104 includes a mounting band 106 that goes around the head and realizes fixing of the device when the user wears it.

The output mechanism 102 includes a housing 108 having a shape that covers the left and right eyes when the head-mounted display 100 is worn by the user, and includes a display panel inside so as to face the eyes when the head-mounted display 100 is worn. Further, a lens located between the display panel and the user's eyes when the head-mounted display 100 is attached may be provided inside the housing 108 to expand the viewing angle of the user. Further, the head-mounted display 100 may further include a speaker or earphone at a position corresponding to the user's ear when worn. Further, the head-mounted display 100 may incorporate various motion sensors such as an acceleration sensor in order to acquire the position and posture of the user's head.

In this example, the head-mounted display 100 includes a stereo camera 110 on the front surface of the housing 108 as the image pickup device 12, and photographs the surrounding real space at a predetermined frame rate in a field of view corresponding to the user's line of sight. According to such a head-mounted display 100, it is possible to acquire the apparent shape and position information of a real object in the user's field of view. If SLAM (Simultaneous Localization and Mapping) technology is introduced, the position and posture of the user's head can be acquired based on the information.

By using such information to determine the field of view for the virtual world, generate display images for left-eye vision and right-eye vision, and display them in the left and right areas of the head-mounted display, it is as if the virtual world is in front of you. It is possible to realize a virtual reality that seems to be spreading. Augmented reality can be realized by superimposing and displaying a virtual object that interacts with a real object that is the subject on a color image taken from the left and right viewpoints. When a color image is used for display and an image of a specific wavelength band is used to obtain information on the subject, the head cloak display 100 may be provided with a plurality of sets of stereo cameras 110 for each wavelength band to be detected.

The information processing device 10 may be an external device capable of establishing communication with the head-mounted display 100, or may be built in the head-mounted display 100. As described above, since the information processing system 1 of the present embodiment can be applied to various aspects, the configuration and appearance shape of each device may be appropriately determined accordingly. In such an aspect, when the position of the subject or the state of the subject space changes, the appearance of the image in the captured image also changes.

For example, even if the subject has a surface shape with many feature points, a clear image may not be obtained in the captured image in an environment with low illuminance. Further, in the embodiment of irradiating light in a specific wavelength band and observing the reflected light, the reflected light from a distant subject may not be obtained with sufficient brightness depending on the irradiation intensity of the light. As a result, it is conceivable that the extraction of the corresponding points in the stereo image fails, the position information cannot be obtained, or the accuracy becomes low.

As a countermeasure against them, it is conceivable to adjust shooting conditions such as exposure time and gain value and image correction parameters, and to adjust the intensity of irradiation light. However, the position of the subject and the degree of light hitting it vary, and even in the same subject space, the optimum conditions are not always one. For example, when adjusting the irradiation intensity of light, by increasing the intensity according to a distant subject, the reflected light may be too strong for a nearby subject and the image may become unclear. Similarly, in a color image, by lengthening the exposure time or increasing the gain value, the entire region that was originally high in brightness may become whitish. In the first place, it is difficult to optimize the combination of shooting conditions, correction parameters, irradiation intensity, etc. each time the situation changes.

Therefore, in the present embodiment, the range of pixel values is amplified by adding the image of the immediately preceding frame to the obtained captured image after keeping these conditions constant. That is, assuming that the imaging device shoots a moving image of a frame at a period of Δt, the same pixel value of the current frame at time t is the same as the frame at the time t−Δt, t-2Δt, ···, t−NΔt. Add the pixel values of the positions. Here, N is a natural number representing the number of past frames to be added. For example, if N = 3, images of 4 frames including the current frame are added.

In this way, the pixel value of the added image is approximately N + 1 times the value of the original image. As a result, the brightness is amplified in the portion where the difference from the surrounding pixels should appear large, such as the reflection position of the irradiation light and the feature point, and the detection becomes easy. By adding the images of another frame, the noise is averaged and the SN ratio becomes high. As a result, even if the distance to the subject is long or the light intensity is low, an image having brightness sufficient for analysis can be obtained, and the position information of the subject can be accurately acquired.

In this way, the brightness of the image can be guaranteed even for a subject having a low intensity of the reflected light, so that it is not necessary to increase the exposure time and the gain value or increase the intensity of the irradiation light. In other words, high-precision analysis is possible without increasing the light intensity on the shooting side. On the other hand, if there is a subject whose brightness is originally obtained, it is not necessary to add the past frames for the subject. Therefore, by providing a processing path for acquiring position information by adding past frames and a processing path for acquiring position information only with the current frame without adding past frames, it is possible to cover various subject states. Each position information can be acquired accurately. Further, by setting the number of past frames to be added to two or more, the luminance amplification factor may be set to three or more steps.

FIG. 3 shows the internal circuit configuration of the information processing device 10. The information processing device 10 includes a CPU (Central Processing Unit) 23, a GPU (Graphics Processing Unit) 24, and a main memory 26. Each of these parts is connected to each other via a bus 30. An input / output interface 28 is further connected to the bus 30. The input / output interface 28 outputs data to a peripheral device interface such as USB or IEEE1394, a communication unit 32 composed of a wired or wireless LAN network interface, a storage unit 34 such as a hard disk drive or a non-volatile memory, and a display device 16. An output unit 36, an input unit 38 for inputting data from an image pickup device 12 or an input device (not shown), and a recording medium drive unit 40 for driving a removable recording medium such as a magnetic disk, an optical disk, or a semiconductor memory are connected.

The CPU 23 controls the entire information processing apparatus 10 by executing the operating system stored in the storage unit 34. The CPU 23 also executes various programs read from the removable recording medium, loaded into the main memory 26, or downloaded via the communication unit 32. The GPU 24 has a geometry engine function and a rendering processor function, performs drawing processing according to a drawing command from the CPU 23, and outputs the drawing process to the output unit 36. The main memory 26 is composed of a RAM (Random Access Memory) and stores programs and data required for processing.

FIG. 4 shows the configuration of the functional block of the information processing device 10. Each functional block of the apparatus shown in FIG. 4 and FIG. 9 to be described later can be realized by various circuits shown in FIG. 3 in terms of hardware, and in terms of software, image analysis loaded from a recording medium into the main memory. It is realized by a program that exerts various functions such as functions, information processing functions, image drawing functions, and data input / output functions. Therefore, it is understood by those skilled in the art that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof, and is not limited to any of them.

The information processing device 10 includes an image data acquisition unit 52 that acquires image data from the image pickup device 12, an image data storage unit 54 that stores the acquired image data, and an image addition unit 56 that adds a predetermined number of images of past frames. Includes an image analysis unit 58 that analyzes an image to obtain position information of a subject, an information processing unit 60 that processes information using analysis results such as position information, and an output unit 62 that outputs data to be output.

The image data acquisition unit 52 is realized by the input unit 38, the CPU 23, the main memory 26, and the like in FIG. 3, and sequentially acquires image data to be captured by the image pickup apparatus 12 at a predetermined frame rate. The data includes data of an image to be detected as a feature point, such as a color stereo image of visible light and a stereo image of reflected light obtained as a result of irradiating light in a specific wavelength band such as infrared light. The image data acquisition unit 52 sequentially stores image data sent at a predetermined frame rate in the image data storage unit 54. The image data storage unit 54 stores at least image data of the current frame and a predetermined number of past frames.

The image addition unit 56 is realized by the CPU 23, GPU 24, main memory 26, etc. of FIG. 3, and is the image data of the current frame stored in the image data storage unit 54 and a predetermined number of past frames stored immediately before the image data. Read the image data. Then, an added image in which the pixel values are added to the pixels at the same position is generated for each of the left and right viewpoints of the stereo image. As described above, a plurality of pairs of added images with different numbers of frames to be added may be generated. The frequency of generating the additive image is determined based on the time resolution required for the position information, and may be the same as or less than the frame rate at which the image pickup apparatus 12 captures the image.

The image analysis unit 58 is realized by the CPU 23, GPU 24, main memory 26, etc. in FIG. 3, and each time the image addition unit 56 generates a stereo image of the addition image, a corresponding point is specified from them, and triangulation is performed based on the parallax. Obtain the distance to the subject based on the principle of. In parallel with this, the image analysis unit 58 identifies the corresponding points from the stereo image of the current frame before addition, and acquires the distance of the subject based on the corresponding points. Then, the image analysis unit 58 integrates the results of both and generates final position information having uniform accuracy regardless of the state of the subject at a predetermined frequency.

The information processing unit 60 is realized by the CPU 23, the main memory 26, and the like in FIG. 3, sequentially acquires the position information generated by the image analysis unit 58, and performs predetermined information processing using the position information. As described above, the content of the information processing performed here is not particularly limited. As a result of the information processing, the information processing unit 60 generates output data such as display images and sounds at a predetermined frequency. At this time, if necessary, the image data of the current frame stored in the image data storage unit 54 is read out and used for generating output data. The output unit 62 is composed of the CPU 23, the output unit 36, and the like shown in FIG. 3, and sequentially outputs the generated output data at a timing suitable for the display device 16.

FIG. 5 is a diagram for explaining a method of irradiating a pattern of infrared rays as an example of the present embodiment and acquiring a distance of a subject by using a photographed image of the reflected light. (A) and (b) of FIG. 5 schematically show captured images of the left viewpoint and the right viewpoint, respectively. Two people 70 and 72 who are the subjects are shown in each photographed image. The person 70 is closer to the image pickup device 12 than the person 72. Further, in the image of the left viewpoint shown in (a), the image of the subject is closer to the right than the image of the right viewpoint shown in (b).

In such a situation, when the subject space is irradiated with spot-shaped infrared rays in a predetermined distribution, the reflected light on the surfaces of the persons 70 and 72 appears in spots on the captured image (for example, images 74a, 74b, 76a, 76b). ). Since the distribution pattern of the irradiated infrared rays is known, the corresponding points in the images of the left and right viewpoints can be obtained based on the distribution pattern of the image of the reflected light in the captured image. For example, it can be seen that the images 74a and 76a in the left viewpoint image of (a) correspond to the images 74b and 76b in the right viewpoint image of (b), respectively.

Here, assuming that the images 74a and 74b representing the same reflection on the person 70 are detected at the lateral positions x1_L and x1_R of the respective images, the parallax is x1_L-x1_R. Similarly, assuming that the images 76a and 76b representing the same reflection on the person 72 are detected at the lateral positions x2_L and x2_R of the respective images, the parallax is x2_L-x2_R. Since the distance to the subject is qualitatively inversely proportional to the parallax, the distance can be derived based on the parallax by obtaining the inverse proportional constant by calibration.

In the illustrated example, since the person 70 is closer, the parallax x1_L-x1_R is larger than the parallax x2_L-x2_R of the person 72. In such a system, when the image of the reflected light in the person 70 near the image pickup device 12, for example, the image of the reflected light in the person 72 is irradiated with infrared rays at an intensity such that the brightness of the images 74a and 74b can be appropriately obtained, the image of the reflected light in the person 72. For example, it is conceivable that the brightness of the images 76a and 76b cannot be obtained, and in some cases, the difference from the noise disappears. In this case, the images 76a and 76b cannot be detected from the image, and the position information of the person 72 cannot be obtained.

FIG. 6 is a diagram for explaining the action of adding frames in the same situation as in FIG. FIG. 6A schematically shows a processing path for generating position information using only the current frame, and FIG. 6B schematically shows a processing path for generating position information using an additive image. Here, the position information is a depth image in which the distance to the subject is a pixel value, and the smaller the distance, the higher the brightness. However, the purpose of location information is not limited to this. In the processing path (a), the corresponding points are detected using only the stereo images 78a and 78b of the current frame, and the distance of the subject is derived from the parallax.

However, when the intensity of the light reflected by the person 72 behind is weak and the image of the reflected light that should be originally does not appear clearly in the image as shown in the figure, accurate parallax is not required and the accuracy of the distance is lowered. As a result, in the depth image 80 generated by the position information acquisition unit 58, the distance value of the front person 70 is accurately represented, while the distance value of the rear person 72 is indefinite and is not represented or contains a lot of errors. The distance value is expressed.

In the processing path (b), the image addition unit 56 first adds the images 82a and 82b of the immediately preceding three frames to the stereo images 78a and 78b of the current frame, respectively, to generate the addition stereo images 84a and 84b. In this addition stereo image, the brightness value is about four times that of the original image, so that the image of the reflected light (for example, image 86) of the person 72, which was not clear before the addition, becomes clear. On the other hand, the image of the reflected light on the person 70 in which the appropriate brightness was obtained in the image before the addition may exceed the upper limit of the brightness assumed in the calculation by the addition.

In the figure, such an image of reflected light (for example, image 88) is shown in a star shape. As a result, in the depth image 90 generated by the position information acquisition unit 58, the distance value of the person 72 behind is accurately represented, and the distance value of the person 70 in front is indefinite and is not represented, or is a distance value including many errors. Is expressed. As described above, the necessity of addition and the appropriate number of frames to be added vary depending on the position of the subject, the intensity of the light hitting the subject, the shooting conditions such as the shooting exposure time, the image correction parameters, and the like.

If multiple processing paths are provided as shown in the figure, analysis is performed independently, and position information is generated for each, regardless of the state of the subject space, irradiation intensity, shooting conditions, correction parameters, etc. Even if you do not adjust, one of them will contain highly accurate information. Therefore, the position information acquisition unit 58 extracts and synthesizes information that is expected to have high accuracy from each position information, and generates one final position information.

FIG. 7 shows how the two depth images 80 and 90 shown in FIG. 6 are integrated. In the illustrated example, the subject whose distance value is represented by the depth image is clearly separated. In such a case, both depth images may be compared, a region in which the distance value is represented in only one of them may be extracted, and the pixel value of the corresponding region in the other depth image may be replaced.

For example, the reference depth image 80 is scanned in raster order or the like, and when a pixel in which a valid value as a distance value is not stored is detected, the same pixel of the other depth image 90 is referred to. If a valid value is stored in the pixel, the pixel value of the original depth image 80 is updated with that value. By performing this process on all the pixels of the depth image 80, it is possible to generate a depth image 92 in which the two depth images 80 and 90 are integrated.

When generating the original depth images 80 and 90, the reliability of the distance value may be acquired for each pixel, and the distance value having the higher reliability may be adopted at the time of integration. For example, in the block matching method in which the position where the highest similarity is obtained by moving the other minute block in the horizontal direction with respect to the minute block defined on one side of the stereo image is obtained as the corresponding point, the corresponding point is determined. The reliability can be determined based on the degree of similarity of the blocks. It is understood by those skilled in the art that the reliability of the distance value can be obtained by various other criteria.

6 and 7 show an example of generating a depth image from a stereo image by a plurality of processing paths, but as described above, in the present embodiment, the information to be output is limited if the processing for detecting the feature point is included. The same effect can be obtained. That is, if the analysis results of a plurality of subjects having different brightness ranges are independently acquired by different processing paths and integrated, the analysis results whose accuracy is guaranteed can be output regardless of the situation of the subject.

Next, the operation of the information processing apparatus according to the present embodiment realized by the above configuration will be described. FIG. 8 is a flowchart showing a processing procedure in which the information processing apparatus 10 acquires position information using captured images and outputs data. This flowchart shows the operation of the information processing device 10 in a state where the user selects electronic content that requires the position information of the subject and the image pickup device 12 is taking a picture. At this time, an initial image is displayed on the display device 16 as needed.

First, the image data acquisition unit 52 of the information processing device 10 acquires stereo image data of the frame at the current time t from the image pickup device 12 and stores it in the image data storage unit 54 (S10). As described above, when the position information is acquired from the stereo image in which a specific wavelength band is detected, the data of the color image may be further acquired. The position information acquisition unit 58 generates a depth image by reading the stereo image of the frame at the time t from the image data storage unit 54, detecting the corresponding points thereof, and obtaining the distance value of the subject (S12).

On the other hand, the image addition unit 56 reads out a predetermined number of stereo image data of the past frame acquired immediately before the current frame from the image data storage unit 54, and adds the pixel values to the pixels at the same position together with the image of the current frame. As a result, the data of the added stereo image is generated (S14). However, as a matter of course, this process is performed from the time when a predetermined number of images of the past frame are stored in the image data storage unit 54. For the number of past frames to be added, find the optimum value logically or experimentally. For example, from the combination of the assumed range of the distance of the subject and the assumed range of the intensity of the irradiation light and the illuminance of the ambient light, the assumed range of the brightness obtained in the stereo image can be found.

An appropriate addition frame is calculated by calculating a magnification such that the brightness value near the lowest value in the assumed range becomes a value sufficient for detecting the corresponding point and sufficiently lower than the upper limit value assumed for the corresponding point acquisition process. You can find the number. Typically, by adding three past frames, the luminance range is quadrupled and the luminance gradation is increased by 2 bits. There may be a plurality of types of addition numbers such as 2, 4, 8, ..., And the processing of S14 and S16 may be performed with the respective addition numbers.

As shown in the added stereo images 84a and 84b of FIG. 6, an image in which sufficient brightness is originally obtained may exceed the upper limit assumed in the calculation by adding. Therefore, the upper limit value of the brightness is set in advance, and when the addition result exceeds the upper limit value, the image addition unit 56 replaces the obtained brightness with the upper limit value. As a result, the present embodiment can be easily introduced without changing the arithmetic program related to the detection of the corresponding points. The position information acquisition unit 58 acquires a stereo image to which past frames have been added from the image addition unit 56, detects corresponding points thereof, and obtains a distance value of the subject to generate a depth image based on the added image. (S16).

Next, the position information acquisition unit 58 integrates the depth image based on the image of the current frame generated in S12 and the depth image based on the additive image to generate one depth image covering various states of the subject ((). S18). The information processing unit 60 performs predetermined information processing using the depth image (S20). As described above, the content of information processing is not particularly limited and may differ depending on the application selected by the user and the like. The information processing unit 60 generates output data such as a display image and sound as a result of information processing, and the output unit 62 outputs the output data to the display device 16 or the like, so that the output corresponds to the situation in the real space at time t. Is done (S22).

While it is not necessary to end the process due to the progress of the game or user operation (N in S24), the image data of the frame at the next time t = t + Δt is acquired (S26, S10), and the processes in S12 to S22 are performed. repeat. When it becomes necessary to end the processing, all the processing is terminated (Y in S24).

According to the present embodiment described above, in the technique of analyzing the captured image to detect the feature points and acquiring the position information of the subject, a predetermined number of images of the past frames captured immediately before the current frame are added. Then perform the analysis. As a result, the brightness range of the pixel, which has poor brightness and is difficult to detect as a feature point, can be amplified, and the noise level can be relatively lowered. As a result, the detection accuracy of the reflected light image and the feature points is improved, and the position information can be accurately obtained.

In addition, a processing path for analysis without adding past frames is provided so as to correspond to the case where appropriate brightness is originally obtained. Alternatively, a processing path is provided in which the number of past frames to be added is different and image analysis is performed for each. By one or a combination of these, the detection accuracy can be guaranteed regardless of the brightness level of the reflected light image or the feature point in the original captured image. As a result, the position of the subject in a wide range of distance can be accurately acquired without adjusting the light irradiation intensity and the shooting conditions.

In addition, a plurality of position information acquired by different processing paths are integrated to generate one depth image. As a result, even if subjects with greatly different distances and states coexist in the same field of view, the accuracy of the finally generated position information becomes uniform, and it is not necessary to consider the variation in accuracy in information processing using the same field of view. As a result, the accuracy of information processing can be easily maintained regardless of the situation of the subject.

Embodiment 2
In the first embodiment, the captured image is used as the target for extracting the feature points, and the past frames are added mainly for the purpose of improving the extraction accuracy. In the present embodiment, images are added for the purpose of expanding the brightness range of the displayed image in a mode in which the captured image is displayed as it is or the captured image is processed and displayed.

At this time, the image of the past frame is corrected in consideration of the movement of the imaging surface and the movement of the subject, and the image of the time of the current frame is generated and then added to make the image of the added image clear. .. The configuration of the information processing system and the internal circuit configuration of the information processing apparatus of the present embodiment may be the same as those described in the first embodiment. Further, the image pickup device 12 and the display device 16 may be configured by the head-mounted display 100 as shown in FIG.

FIG. 9 shows the configuration of the functional block of the information processing device according to the present embodiment. The information processing device 150 includes an image data acquisition unit 152 that acquires image data from the image pickup device 12, an image data storage unit 154 that stores the acquired image data, and an image addition unit 156 that adds a predetermined number of images of past frames. Includes a state information acquisition unit 158 that acquires information related to the state of the subject, and an output unit 62 that outputs data to be output.

The image data acquisition unit 152 has the same function as the image data acquisition unit 52 of the first embodiment. However, the image data acquisition unit 152 in the present embodiment may acquire at least the data used for the display image. The image data acquisition unit 152 further acquires data for obtaining changes in the relative positions and postures of the imaging surface and the subject in association with the captured image. For example, when the head-mounted display 100 shown in FIG. 2 is introduced and a display is performed using an image taken by a camera provided on the front surface of the housing 108, the motion of the gyro sensor, acceleration sensor, etc. built in the head-mounted display 100 is performed. By acquiring the measured value from the sensor, the movement of the user's head is required.

As a result, the relative movement of the subject with respect to the captured image plane can be specified, so that the image of the subject reflected in the image of the past frame can be corrected to the image at the same time as the current frame. If the relative movement of the subject with respect to the captured image plane can be specified, the data on which it is based is not limited to the measured value of the motion sensor, and the image pickup device 12 and the display device 16 of the present embodiment are limited to the head-mounted display 100. Not the point.

For example, when the subject has a known shape and size, its position and posture in the real space can be obtained by using a photographed image by matching with a template image or an object model. In addition, various techniques for tracking or estimating changes in the position or posture of a subject in real space using captured images have been proposed, and any of them may be applied.

The image data acquisition unit 152 sequentially stores image data sent at a predetermined frame rate in the image data storage unit 154. The image data storage unit 154 stores image data of the current frame and a predetermined number of past frames. When acquiring the measured value of the motion sensor from the head-mounted display 100, the image data acquisition unit 152 sequentially stores the data in the image data storage unit 154 in association with the captured image at each time.

The state information acquisition unit 158 is realized by the CPU 23, GPU 24, main memory 26, etc. in FIG. 3, and sequentially reads out the measured values of the motion sensor or the data of the captured image stored in the image data storage unit 154, and three-dimensionally as described above. Acquires the position and orientation of the subject in real space for each time. The acquired information is sequentially stored in the image data storage unit 154 in association with the captured image at each time. The image addition unit 156 is realized by the CPU 23, GPU 24, main memory 26, etc. of FIG. 3, and includes a correction unit 164 and an addition unit 166.

The correction unit 164 acquires the rotation angle and translation amount generated from the past frame to the current frame for each subject based on the information on the position and posture of the subject in each frame acquired by the state information acquisition unit 158. Then, by virtually manipulating the position and orientation of the subject in the three-dimensional space, the image when the subject in the image of the past frame is advanced to the time of the current frame is obtained.

The addition unit 166 generates an addition image to be used for display by adding the image of the past frame corrected in this way to the image of the current frame. The number of frames to be added is determined based on the brightness range of the original captured image and the brightness range supported by the display device 16 or the brightness range desired for image expression. The number of frames to be added may be adaptively determined according to the connected display device 16.

The output unit 162 includes the CPU 23 and the output unit 36 of FIG. 3, and sequentially outputs the data of the added image generated by the image adding unit 156 at a timing suitable for the display device 16. The output unit 162 may output data after performing predetermined processing such as drawing a virtual object on the added image. In such processing, the information related to the position and posture of the subject acquired by the state information acquisition unit 158 may be used. The output unit 162 may also output audio data.

FIG. 10 schematically shows how the image addition unit 156 in the present embodiment corrects the image of the past frame and then adds it to the image of the current frame. The upper part of the figure illustrates the shooting times of four frames to be added and the contents of each shot image when the shooting cycle of each frame is Δt and the shooting time of the current frame is t. In the example of the figure, the disk gradually changes from the state where the side surface of the disk-shaped object which is the subject is visible at the time t-3Δt to the time t-2Δt, t−Δt, t, and so on. It has changed so that the top surface can be seen. The subject is moving from left to right in the field of view.

At this time, the state information acquisition unit 158 acquires information on the position and posture of the subject in the three-dimensional space at each time, or the amount of change thereof, as shown below each captured image. The correction unit 164 corrects the image of the subject in the past frame to the time state of the current frame based on the difference between the positions and orientations of the past frame and the current frame, that is, the rotation angle and the translation amount on the three axes. Assuming that the angle of rotation of the subject is roll φ, pitch θ, and yaw ψ, and the translation amount is (T _x , T _y , T _z ), the surface of the subject that matches the position coordinates (x, y, z) in the three-dimensional space. The point moves to the position coordinates (x', y', z') after rotation and translation according to the following equation.

If the position coordinates (x', y', z') obtained by the above equation are projected onto the image plane by fluoroscopic transformation, the movement destination of the pixels in the original captured image can be obtained. By performing this correction process on all the pixels that make up the subject, as shown in the middle part of the figure, from the image of the subject in each frame at time t-3Δt, t-2Δt, and t−Δt, the current frame An image at time t can be generated. The addition unit 166 adds the image corrected by the image of the past frame at time t-3Δt, t-2Δt, and t−Δt and the captured image of the current frame at time t, and after the addition shown in the lower part of the figure. Image can be generated. The color depth of this image is increased by 2 bits from the original captured image. Therefore, if the display device 16 corresponding to this is used for display, more dynamic image expression becomes possible.

Next, the operation of the information processing apparatus according to the present embodiment realized by the above configuration will be described. FIG. 11 is a flowchart showing a processing procedure in which the information processing device 150 expands the brightness range of the captured image and displays it. This flowchart shows the operation of the information processing device 150 in a state where the user selects an electronic content accompanied by a display using a captured image and the image pickup device 12 captures the image. At this time, an initial image is displayed on the display device 16 as needed.

First, the image data acquisition unit 152 of the information processing device 150 acquires the data of the captured image of the frame at the current time t from the image pickup device 12 and stores it in the image data storage unit 54 (S30). At this time, depending on the mode, the measured value related to the position and posture of the user's head is acquired from the motion sensor of the head-mounted display 100 that also serves as the image pickup device 12, and is associated with the captured image data in the image data storage unit 54. Store.

The state information acquisition unit 158 acquires the position and orientation of the subject in the frame at the current time t (S32). Since the purpose of acquiring the information is to correct the image on the image plane as shown in FIG. 10, the state information acquisition unit 158 obtains information capable of deriving the relative position and angle relationship between the imaging surface and the subject. get. To that extent, the information may be the position and orientation of the subject and screen in the world coordinate system, or the position and orientation of the subject in the camera coordinate system.

As described above, such information may be acquired from the measured value by the motion sensor of the head-mounted display 100, or may be acquired based on the shape and size of the image of the subject appearing in the captured image. The stereo image data may be acquired from the image pickup apparatus 12, and the position of the subject in the three-dimensional space may be specified based on the data. In this case, as described in the first embodiment, the corresponding point may be detected by using an image obtained by adding the captured images of the past frames. The information based on the motion sensor and the information based on the captured image may be integrated to obtain the final position and posture information.

The acquired information is stored in the image data storage unit 154 in association with the data of the captured image of the frame at time t. Subsequently, the correction unit 164 of the image addition unit 156 stores the data of the captured images of a predetermined number of past frames acquired immediately before the current frame and the information of the position and orientation of the subject associated with them as the image data storage unit. Read from 54 and generate an image in which the image of the subject is corrected to the state of the current time t (S34).

Specifically, the subject is selected from the rotation angle and translation amount of the subject generated from the shooting time t−nΔt of the past frame (1 ≦ n ≦ N, N is the number of past frames to be added) to the shooting time t of the current frame. The position coordinates of the movement destination of each of the constituent pixels are obtained by the above equation and perspective conversion. Then, by moving the pixels of the image before correction, the image after correction is formed. If it is considered that the ratio of the error to the translation amount of the object in several frames is large, the translation amount components (T _x , T _y , T _z ) in the above equation need not be included in the calculation.

Further, the correction unit 164 may perform further correction processing in order to further improve the image quality. Specifically, the pixels of the images of the past frame and the current frame may be interpolated by an existing method to increase the resolution. Further, various correction filters such as a noise removal filter may be applied. The addition unit 166 generates an addition image by adding the image corrected in this way to the pixels at the same position together with the image of the current frame (S36). As will be described later, high-definition and high-resolution of the added image may be realized by shifting the area of the pixels to be added in units of sub-pixels.

As in the first embodiment, when a pixel whose luminance value exceeds a predetermined upper limit is generated by addition, the pixel value is replaced with the upper limit value. The output unit 162 outputs the generated data of the added image to the display device 16 or the like (S38). As a result, the display image at time t is displayed in high definition. The output unit 162 may also output audio data as appropriate. Further, as described above, the output unit 162 may perform a predetermined process on the added image.

While it is not necessary to end the process due to the progress of the game or user operation (N in S40), the captured image data of the frame at the next time t = t + Δt is acquired (S42, S30), and the processes in S32 to S38. repeat. When it becomes necessary to end the processing, all the processing is terminated (Y in S40).

FIG. 12 schematically shows how the addition unit 166 shifts the pixel area and adds images in S36 of FIG. In this example, the positional deviation of the pixels for 2 rows and 3 columns when adding the two images 170 and 172 is illustrated, the boundary line of the pixels of the image 170 is a solid line, and the boundary line of the pixels of the image 172 is a broken line. It is shown by. For example, pixel 174 in image 170 corresponds to pixel 176 in image 172. When the pixel boundary is shifted by half a pixel in both the vertical and horizontal directions as shown in the figure, the pixel area of one image is divided into four by the boundary of the pixel area of the other image.

For example, pixel 176 is divided into four regions of "A", "B", "C", and "D". The region "A" is added to the pixel value of pixel 174 of the other image 170. Each of the other regions is added to the pixel value of a different pixel of the image 170 adjacent to the pixel 174. As a result, the resolution of the image obtained by adding both is four times that of the original image. By this processing, a high-resolution image with higher accuracy can be generated as compared with linear interpolation between pixels.

In the illustrated example, the vertical and horizontal shift amounts are the same, but in the present embodiment, the rotation angle and translation amount of the subject are obtained, so that the shift amount is made anisotropic accordingly. May be good. For example, the pixel movement amount before and after the correction may be shifted larger in the larger direction. Further, especially when adding images of 3 frames or more, the shift amount is not limited to half a pixel, and may be a unit smaller than that. That is, a rule may be set in advance so that the combination of the direction and amount of shifting each image at the time of addition can be derived from the number of frames to be added and the position and orientation of the subject.

According to the present embodiment described above, in the technique of displaying using captured images, a predetermined number of images of past frames captured immediately before the current frame are added to generate a displayed image. As a result, the brightness of the image of the subject can be controlled without amplifying the noise, so that a more expressive and high-definition image can be displayed according to the brightness range that the display device can handle.

Further, the position and posture of the subject in the three-dimensional space are acquired at each time, and the image in the image of the past frame before addition is corrected according to the time of the current frame. As a result, even if the imaging surface or the subject moves due to the passage of time from the past frame, a clear image can be displayed without affecting the minute movement. As a result, the number of past frames to be added can be freely set while minimizing the influence on the displayed image, and conversion to a desired brightness range can be easily realized.

The function of the correction unit 164 of the present embodiment may be provided in the image addition unit 56 of the first embodiment. In this case, the information processing device 10 may further be provided with a state information acquisition unit 158 that acquires the rotation angle and translation amount of the subject in the three-dimensional space based on the measured values of the motion sensor, or the image analysis unit 58 may provide the information processing unit 58. The process may be performed and the result may be supplied to the image addition unit 56. As a result, an additive image can be generated in consideration of minute movements of the subject, feature points can be detected, and position information can be acquired with high accuracy. In addition, since the degree of freedom for the number of past frames to be added increases, it is possible to accurately acquire the feature points of subjects in more diverse situations.

The present invention has been described above based on the embodiments. The above-described embodiment is an example, and it is understood by those skilled in the art that various modifications are possible for each of these components and combinations of each processing process, and that such modifications are also within the scope of the present invention. be.

For example, in the first embodiment, the image analysis unit 58 independently acquires the position information in the processing path in which the captured image is not added and the processing path in which the captured image is added, and integrates the results. On the other hand, the image analysis unit 58 may separate the processing paths from other viewpoints, acquire position information for each, and integrate the results. For example, a path for acquiring position information using a color stereo image obtained by photographing ambient light and a path for acquiring position information using a stereo image obtained by photographing light in a specific wavelength band may be provided.

Then, in each processing path, the captured image may be further branched into processing paths having different addition / non-addition and addition numbers. As the number of processing paths increases, feature point detection and position information acquisition can be performed with higher robustness against changes in the subject's situation. The processing path setting is appropriately determined according to the accuracy and time resolution required for the acquired information, the processing performance of the information processing device, the allowable communication band, and the like.

1 Information processing system, 10 Information processing device, 12 Imaging device, 16 Display device, 52 Image data acquisition unit, 54 Image data storage unit, 56 Image addition unit, 58 Image analysis unit, 60 Information processing unit, 62 Output unit, 152 Image data acquisition unit, 154 image data storage unit, 156 image addition unit, 158 status information acquisition unit, 162 output unit, 164 correction unit, 166 addition unit.

Claims (8)

An image data acquisition unit that sequentially acquires image data of frames of moving images to be captured, and an image data acquisition unit.
An image addition unit that generates an addition image by adding the pixel value of the image of the past frame acquired before that to the pixel value of the image of the newly acquired current frame at the corresponding position.
An image analysis unit that extracts feature points from the added image and performs a predetermined analysis process, extracts feature points from the image of the current frame and performs the same analysis process, and integrates the results of both.
An output unit that outputs data representing the integrated result, and
An information processing device characterized by being equipped with. The image data acquisition unit acquires stereo image data obtained by capturing the same space from the left and right viewpoints as the image data.
The image analysis unit obtains depth images representing the distance to the subject based on the stereo image of the added image and the feature points extracted from the stereo image of the current frame, and integrates the two to obtain different distances. The information processing apparatus according to claim 1, wherein one depth image relating to a subject is generated. The image addition unit generates a plurality of the addition images in which the number of the past frames to be added is different.
The information processing apparatus according to claim 1 or 2, wherein the image analysis unit performs analysis processing based on feature points extracted from the plurality of added images. The information processing apparatus according to claim 2 or 3, wherein the image adding unit determines the number of past frames to be added based on at least an assumed range of a distance to a subject. The image data acquisition unit acquires image data of a frame in which the reflected light of light in a predetermined wavelength band irradiated in the subject space is captured as a moving image.
The information processing apparatus according to any one of claims 1 to 4, wherein the image analysis unit extracts an image of the reflected light from the added image as the feature point. A head-mounted display equipped with an imaging device that captures moving images in a field of view that corresponds to the user's line of sight,
An information processing device that generates data for a display image to be displayed on the head-mounted display based on the moving image.
With
The information processing device
An image data acquisition unit that sequentially acquires image data of frames of the moving image,
An image addition unit that generates an addition image by adding the pixel value of the image of the past frame acquired before that to the pixel value of the image of the newly acquired current frame at the corresponding position.
An image analysis unit that extracts feature points from the added image and performs a predetermined analysis process, extracts feature points from the image of the current frame and performs the same analysis process, and integrates the results of both. ,
An output unit that generates and outputs the data of the display image using the integrated result,
An information processing system characterized by being equipped with. The step of sequentially acquiring the image data of the frame of the moving image to be captured and storing it in the memory,
A step of generating an added image by adding the pixel values of the newly acquired image of the current frame to the pixel values of the image of the past frame acquired earlier than that read from the memory at the corresponding positions.
A step of extracting feature points from the added image and performing a predetermined analysis process, extracting feature points from the image of the current frame and performing the same analysis process, and integrating the results of both.
Steps to output data representing the result of integration,
An image processing method by an information processing apparatus, which comprises. A function to sequentially acquire image data of frames of moving images to be shot, and
A function to generate an added image by adding the pixel value of the image of the past frame acquired before that to the pixel value of the image of the newly acquired current frame at the corresponding position.
A function of extracting feature points from the added image and performing a predetermined analysis process, extracting feature points from the image of the current frame and performing the same analysis process, and integrating the results of both.
A function to output data representing the integrated result and
A computer program characterized by realizing a computer.

Images