JP2009212664A - Encoding method, decoding method, encoding device, decoding device, encoding program and decoding program, of distance information, and computer readable recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To encode distance information indicating the distance from a camera to a subject more efficiently than before by performing suitable conversion on distance information used as reference. <P>SOLUTION: A reference information conversion unit 106 calculates three-dimensional coordinates of the subject that the reference distance information obtained by decoding distance information used to predict distance information to be encoded and having been encoded indicates, using the reference distance information and reference camera information, and obtains the distance from the camera indicated by camera information to be encoded to the point of the three-dimensional coordinates. A distance information encoding unit 108 quantizes the distance and uses the quantized converted reference distance information as a reference object to predictively encodes the distance information to be encoded. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a technique for encoding and decoding multi-view distance information.

  A multi-view image is a plurality of images obtained by photographing the same subject and background with a plurality of cameras, and a multi-view video (multi-view video) is a moving image. The distance information referred to here is information representing the distance from the camera to the subject for each region given to a certain image. The multi-view distance information is distance information for a multi-view image, and is a set of a plurality of normal distance information. Since the distance from the camera to the subject can be called the depth of the scene, the distance information is sometimes called depth information.

  In general, such distance information is given to a two-dimensional plane obtained as a result of being photographed by a camera, so that the distance information is represented as a distance image by mapping the distance to a pixel value of the image. Since the information for a certain point on the two-dimensional plane is only one distance, the distance image can be expressed as a gray scale image. The distance image is sometimes called a depth image or a depth map.

  One of the uses of distance information is a stereoscopic image. In general stereo image representation, a stereo image is composed of an observer's right-eye image and left-eye image, but a stereo image can be represented using an image from a camera and its distance information ( For details, see Non-Patent Document 1.)

  As a method for encoding a stereoscopic video represented by using the video at one viewpoint and the distance information, MPEG-C Part. 3 (ISO / IEC 23002-3) can be used (refer to Non-Patent Document 2 for details).

  The multi-view distance information is used to represent a stereoscopic image having a larger parallax than a stereoscopic image that can be expressed using single-view distance information (see Non-Patent Document 3 for details).

  In addition to the purpose of representing such stereoscopic images, multi-view distance information can be used as one of data for generating a free viewpoint image that allows the viewer to freely move the viewpoint without worrying about the location of the shooting camera. used. A composite image when a scene is viewed from a camera different from such a camera is sometimes referred to as a virtual viewpoint image, and its generation method is actively studied in the field of Image-based Rendering. As a representative method for generating a virtual viewpoint video from multi-view video and multi-view distance information, there is a method described in Non-Patent Document 4.

  As described above, the distance information can be regarded as a gray scale moving image, and the subject exists continuously in the real space and cannot move instantaneously. It can be said that there is a correlation. Therefore, distance information is efficiently encoded while removing spatial redundancy and temporal redundancy by an image encoding method and a moving image encoding method used for encoding a normal video signal. Actually MPEG-C Part. In No. 3, encoding is performed using an existing moving image encoding method.

  Here, a conventional general video signal encoding method will be described. In general, since an object has spatial and temporal continuity in real space, its appearance is highly correlated in space and time. In video signal encoding, such a correlation is utilized to achieve high encoding efficiency.

Specifically, the video signal of the encoding target block is predicted from the already encoded video signal, and only the prediction residual is encoded, thereby reducing the information that needs to be encoded, Achieve efficiency. As typical video signal prediction methods, in single-view video, intra-screen prediction that generates a prediction signal spatially from adjacent blocks, and estimation of subject motion from encoded frames taken at close times Then, there is motion compensated prediction that generates a prediction signal in time, and in multi-view images, in addition to these, the parallax of the subject is estimated from an encoded frame taken by another camera, and the prediction signal is transmitted between the cameras. There is a disparity compensation prediction that generates Details of each method are described in Non-Patent Document 5, Non-Patent Document 6, and the like.
C.Fehn, P.Kauff, M.Op de Beeck, F.Emst, W.IJsselsteijn, M.Pollefeys, L.Van Gool, E.Ofek and I.Sexton, “An Evolutionary and Optimised Approach on 3D-TV” , Proceedings of International Broadcast Conference, pp. 357-365, Amsterdam, The Netherlands, September 2002. WHA Bruls, C. Varekamp, R. Klein Gunnewiek, B. Barenbrug and A. Bourge, “Enabling Introduction of Stereoscopic (3D) Video: Formats and Compression Standards”, Proceedings of IEEE International Conference on Image Processing, pp.I-89 -I-92, San Antonio, USA, September 2007. A.Smolic, K.Mueller, P.Merkle, N.Atzpadin, C.Fehn, M.Mueller, O.Schreer, R.Tanger, P.Kauff and T.Wiegand, “Multi-view video plus depth (MVD) format for advanced 3D video systems ”, Joint Video Team of ISO / IEC JTC1 / SC29 / WG11 and ITU-T SG16 Q.6, Doc. JVT-W100, San Jose, USA, April 2007. CLZitnick, SBKang, M. Uyttendaele, SAJWinder, and R. Szeliski, “High-quality Video View Interpolation Using a Layered Representation”, ACM Transactions on Graphics, vol.23, no.3, pp.600-608, August 2004. ITU-T Rec.H.264 / ISO / IEC 11496-10, “Advanced Video Coding”, Final Committee Draft, Document JVT-E022, September 2002. H. Kimata and M. Kitahara, “Preliminary results on multiple view video coding (3DAV)”, document M10976 MPEG Redmond Meeting, July, 2004.

  The subject has a high spatial correlation because it is continuous in real space, and has a high temporal correlation because it cannot move instantaneously. Therefore, it is possible to efficiently encode distance information represented as a grayscale image by using an existing video encoding method that uses spatial correlation and temporal correlation.

  However, when the position and orientation of the camera are different as shown in FIG. 7, the distance from the camera to the subject is different even for the same subject. In such a case, even if the subject is the same, the distance varies from frame to frame, so it is impossible to accurately predict the distance in time or space.

  In order to respond to changes in the brightness of the entire scene, etc. There is a technique for efficiently predicting a value that changes for each frame by changing the original value of the reference destination and the value to be referred to using H.264 / AVC weighted prediction or the like.

  In such a method, since a single converted value is given for a given value, it becomes a many-to-one conversion. However, as shown in FIG. 8, even if the subject is the same distance from one camera, the subject may be a different distance from another camera. In this case, the conversion needs to handle one-to-many conversion, and the conventional processing cannot perform efficient prediction.

  Also, as shown in FIG. 9, the range of distances to be expressed may differ depending on the camera. When the expression ranges are different in this way, setting a common expression range increases the bit depth necessary for the expression, resulting in a decrease in encoding efficiency. Furthermore, when not only the expression range but also the distance sampling interval is different, creating sampling points that include all results in a significant increase in the required bit depth.

  The present invention has been made in view of such circumstances, and an object of the present invention is to encode distance information more efficiently than before by performing appropriate conversion on distance information used as a reference.

  In order to solve the above-described problems, in the present invention, when encoding certain distance information, the distance information to be referred to is the same distance axis and expression range used to represent the distance information to be encoded. By converting so that it can be expressed by the sampling interval, efficient prediction encoding can be performed without increasing the bit depth even when the encoding target and the reference destination have different cameras.

  By this conversion, a change in value due to the movement of the camera included in the distance information can be removed, and a change in position due to the movement of the camera and a change in position and value due to the movement of the subject need only be predicted and encoded. That is, it is clear that efficient predictive coding can be realized because less information has to be predicted.

  The specific conversion is performed by identifying the three-dimensional position of the observed object from the reference distance information and the camera information, and calculating the distance from the three-dimensional position to the encoding target camera. Is called. Note that this conversion is performed according to the physical phenomenon when an object in real space is photographed by the camera. Therefore, if the projective conversion by the camera can be sufficiently modeled, the conversion can be realized with very high accuracy. Can do.

  Furthermore, a camera that exists between the reference distance information and the encoding target distance information by converting only the value of the reference distance information to a position that is considered to be observed by the encoding target camera. It is possible to remove not only a change in value due to a change in position of the camera but also a change in position due to a change in the position of the camera. As a result, the number of components that must be further predicted is reduced, and more efficient predictive coding can be realized.

  At this time, using the plurality of converted reference distance information, the distance information that takes the average value, the median value, and the value representing the closest to the camera for each position is obtained, and the new distance information is synthesized, You may use as a reference at the time of encoding. When multi-viewpoint distance information includes multiple reference distance information from different cameras at the same time, this process reduces the effects of noise, etc. included in the reference distance information. In addition, the influence of occlusion can be reduced. Occlusion is an area that can be observed on the one hand, but not on the other hand, because the camera is different. Since this occlusion area varies depending on the combination of cameras, it is possible to reduce the occlusion area by integrating distance information for a plurality of cameras.

  Instead of converting only the value of the reference distance information, instead of referring to the converted reference distance information normally when converting to a position that is considered to be observed by the encoding target camera, the value is encoded in advance. By subtracting from the encoding target distance information and encoding only the difference, the encoding target information can be reduced and efficient encoding can be realized.

  In particular, when the multi-viewpoint distance information includes multiple reference distance information at different times at the same time, the component remaining in the difference distance information generated by such subtraction processing is the reference camera. And components generated due to differences in sampling interval and expression range of the encoding target camera and components of the occlusion part. In normal video coding, one prediction method must be selected for each area. Therefore, if such subtraction processing is not performed, the sampling interval and expression range of the reference camera and encoding target camera as in the former will be reduced. In a region where a difference occurs, a prediction error depending on a sampling difference or the like should be encoded if the prediction in the spatial direction is used, and a prediction error due to the movement of the subject if the prediction in the time direction is used.

  On the other hand, by performing the subtraction process, only the value depending on the difference in sampling remains, so it is possible to further reduce the information to be encoded by predicting it in the time direction. Regarding the latter occlusion part, the temporally predictable component remains in the difference as it is, and there is no demerit by performing the subtraction process. Therefore, it is possible to improve the overall encoding efficiency by performing such subtraction processing as much as the encoding efficiency can be increased in the former region.

  Even in this case, a plurality of converted reference distance information is used to obtain distance information that takes an average value, a median value, and a value representing the closestness to the camera for each position. And the combined distance information may be subtracted from the encoding target distance information. In this case, the influence of noise or the like included in the reference distance information can be reduced, and the occlusion area can be reduced. The effect of noise is obviously a positive effect on coding, and the reduction of the occlusion area means that the area where coding efficiency does not improve even if subtraction processing is performed decreases. In order to indicate that the number of areas in which the conversion efficiency is improved increases, the differential distance information can be further efficiently encoded.

  According to the present invention, the distance information of the reference object is appropriately determined even when the position and orientation of the camera, the distance expression range, the sampling interval, and the like serving as the reference of the distance information of the reference object and the distance information of the encoding object are different. By converting to, it becomes possible to encode distance information more efficiently than in the past.

  Hereinafter, the present invention will be described in detail according to embodiments. First, the first embodiment (hereinafter referred to as the first embodiment) will be described. FIG. 1 shows a configuration diagram of a distance information encoding apparatus according to Embodiment 1 of the present invention.

  As shown in FIG. 1, a distance information encoding apparatus 100 includes a distance information input unit 101 that inputs distance information to be encoded, a distance information memory 102 that stores input distance information, and an encoding target. A camera information input unit 103 that inputs camera parameters and the like of a camera that is subject to distance information, and reference distance information that inputs already encoded and decoded distance information that is referred to when predictively encoding distance information to be encoded A reference distance having an input unit 104, a reference camera information input unit 105 for inputting camera parameters and the like of a camera whose reference distance information is a target, and reference three-dimensional point restoration means, and converting reference distance information using this Information conversion unit 106, reference distance information memory 107 for storing the converted reference distance information, and distance accumulated in reference distance information memory 107 And a distance information encoding unit 108 for prediction encoding the distance information stored in the distance information memory 102 with reference to the distribution.

  FIG. 2 shows a processing flow executed by the distance information encoding apparatus 100 configured as described above. The processing executed by the distance information encoding apparatus 100 will be described in detail according to this processing flow.

First, distance information to be encoded is input from the distance information input unit 101 and stored in the distance information memory 102 [step A1]. Hereinafter, the distance information of the encoding target represents a D _in, representing the distance information for a specific region by adding information for identifying the position sandwiched by the symbol [].

Next, the information of the camera parameters such as camera D _in is a reference is input from the camera information input unit 103 [Step A2]. In the following, the camera internal parameter matrix for D _in is represented by A _in , the rotation matrix is represented by R _in , and the translation vector is represented by t _in . Since there are various representations of camera parameters, the mathematical formulas used below need to be changed according to the definition of camera parameters. In this embodiment, it is assumed that the correspondence between the image coordinate m and the world coordinate M uses a camera parameter expression obtained by the expression M ^* = RA ⁻¹ m ^* + t. A, R, and t represent the camera internal parameter matrix, rotation matrix, and translation vector, respectively, and M ^* and m ^* represent homogeneous coordinates that allow arbitrary scalar multiplication of M and m, respectively. A and R are 3 × 3 matrices, and t is a three-dimensional vector.

In this embodiment, it is assumed that each time and distance information of each camera is given as a gray scale image. Information necessary for associating the distance from the camera to the subject with the pixel value is also included in the camera information input in step A2. For example, the information required by the method to associate changes, but the look-up table (Look up table) and, in correspondence of the minimum value MinD _in · maximum MaxD _in · the number of steps STEPd _in such distance and pixel value It becomes necessary information. In the latter case, the calculation formula S _in (d) for quantizing the distance d is expressed by Equation 1 when the distance value itself is uniformly quantized, and when the reciprocal of the distance is uniformly quantized. Can be represented by Equation 2.

  When the input related to the distance information to be encoded is completed, reference distance information used for predictive encoding is prepared [step A3-A8]. The process here is performed for each distance information obtained by decoding the already encoded distance information used as a reference destination. That is, if the index representing the distance information frame as a reference candidate is ref and the number of reference candidate frames is numRefs, ref is initialized to 0 [step A3], and 1 is added to ref [step A7], ref Until step NumRefs is reached [step A8], the following steps A4-A6 are repeated.

In the process performed for each reference candidate, first, reference distance information obtained by decoding already encoded distance information as a reference candidate is input from the reference distance information input unit 104 [step A4]. In the following, the reference distance information of the index ref is represented as D _ref, and the distance information of a specific region is represented by adding information for identifying a position sandwiched between symbols [].

Next, camera information based on D _ref is input from the reference camera information input unit 105 [step A5]. The camera information here includes information necessary for associating the camera parameter and the distance from the camera to the subject with the pixel value, similarly to the camera information for the encoding target camera. In the following, the camera internal parameter matrix for D _ref is represented by A _ref , the rotation matrix is represented by R _ref , the translation vector is represented by t _ref , and the function for quantizing the distance d from the camera to the subject is represented by S _ref (d).

When the reference distance information and the reference camera information are obtained, the reference distance information conversion unit 106 converts the reference distance information D _ref and stores it in the reference distance information memory 107 [Step A6]. Details of the conversion here will be described later.

  When the conversion of all the reference distance information is completed, the distance information encoding unit 108 uses the reference distance information stored in the reference distance information memory 107 as a reference candidate in predictive encoding, and stores it in the distance information memory 102. The encoding target distance information stored is predictively encoded [step A9]. Note that any method may be used for the predictive coding here. For example, MPEG-2 and H.264. The encoding may be performed by a method compliant with an international standard method for moving image encoding such as H.264 / AVC.

FIG. 3 shows a detailed flow of reference distance information conversion processing in step A6 of FIG. In the conversion process, first, compared to the camera the camera parameters for the camera parameters and D _ref camera for D _in is performed [step B1]. Specifically, it is checked whether R _in and R _ref of the rotation matrix are equal and the third component of each translation vector (t _in , t _ref ) is equal. If this condition is satisfied, the camera relative positional relationship with respect to the camera and D _ref for D _in the orientation is the same, results in only a vertical positional shift relative to its orientation, distance from the camera to the subject Does not change. It should be noted that the determinations here do not have to completely match, and may be determined to match if the difference is within a certain range considering camera parameter estimation errors.

If the camera parameter check is satisfied, the quantization function is checked [Step B2]. For example, if quantization is performed using a look-up table, whether the numbers are equal, or if quantization is performed using the minimum value, maximum value, and number of steps , Check whether each value is equal. If it is determined that the same quantization function conversion is not performed at all, the value D _'ref after conversion of D _ref is terminated as D _ref [step B3].

On the other hand, if it is determined that the quantization functions are different, inverse quantization using S _ref and re-quantization using S _in are performed for each sample to generate a value of D ′ _ref [Step B4- B7]. Here, the index of the sample is represented as pix, and the number of samples is represented as numPixs. Specifically, after pix is initialized to 0 [Step B4], 1 is added to pix [Step B6], and until pix becomes numPixs [Step B7], the inverse quantization / re-encoding according to the following equation is performed. Quantization is executed [step B5].

D ′ _ref [pix] = S _in (S _ref ⁻¹ (D _ref [pix])) (Formula 3)
If not satisfied check for camera parameters, to restore the three-dimensional coordinates by performing a back projection using the camera parameters for the D _ref for each sample, the restored three-dimensional coordinate reprojection using the camera parameters for D _in Thus, the distance is converted [step B8-B13]. Here, the index of the sample is represented as pix, and the number of samples is represented as numPixs. Specifically, after initializing pix with 0 [Step B8], adding 1 to pix [Step B12], until pix becomes numPixs [Step B13], the following processing [Step B9-B11] repeat.

In the process repeated for each sample, first, the three-dimensional coordinates are restored by the back projection process [step B9]. The processing here is expressed by the following Expression 4. Note that g _pix represents a three-dimensional coordinate value, and (u _pix , v _pix ) represents the position of pix on the grayscale image.

Next, a process of reprojecting the obtained three-dimensional coordinates is performed [step B10]. The processing here is expressed by the following Equation 5. Note that (x _pix , y _pix , z _pix ) are homogeneous coordinate values in the image coordinate system corresponding to g _pix .

  When backprojection and reprojection are performed according to the above equations 4 and 5 with the above camera parameter definition, the third component on the left side of equation 5 is the distance value after conversion. Since this value is not quantized, the quantization process is finally performed according to the following equation 6 [step B11].

D ′ _ref [pix] = S _in (z _pix ) (Formula 6)
There is also a method of executing the reference distance information conversion process in step A6 according to the flow shown in FIG. Most of the conversion processing is the same as the conversion by the flow shown in FIG. 3 described above. When this flow is followed, D' _ref is first initialized [Step C1]. In this initialization, a value indicating the farthest from the camera is substituted as a value for all pixels.

Next, comparison of the camera the camera parameters for the camera parameters and D _ref camera for D _in is performed [step C2]. Specifically, it is checked whether R _{in of} the rotation matrix is equal to R _ref and whether t _{in of} the translation vector is equal to t _ref . The process [Step C3-C8] when the check for the camera parameter is satisfied is the same as the process [Step B2-B7] of FIG. 3 already described.

If not satisfied check for camera parameters, to restore the three-dimensional coordinates by performing a back projection using the camera parameters for the D _ref for each pixel, the reprojection the restored three-dimensional coordinates by using the camera parameters for D _in The distance is converted by performing [Step C9-C15]. Here, the pixel index is represented as pix, and the number of pixels is represented as numPixs. Specifically, after pix is initialized with 0 [Step C9], 1 is added to pix [Step C14], and the following processing is repeated until pix becomes numPixs [Step C15] [Step C10- C13].

  The first half of the process repeated for each pixel [Step C10-C11] is the same as the process of FIG. 3 [Step B9-B10], and the respective three-dimensional coordinate restoration [Step C10] by back projection processing is obtained. A three-dimensional coordinate reprojection process [step C11] is performed. Each process is expressed by the above-described Expression 4 and Expression 5.

Next, when 3D coordinates were projected to be projected onto the encoding target camera, it was already obtained at the position where the subject was sampled (x _pix / z _pix , y _pix / z _pix ) The distance is compared with the distance obtained by the current process [Step C12]. Specifically, the comparison shown by the following equation 7 is performed.

z _pix <S _in ⁻¹ (D ′ _ref [x _pix / z _pix , y _pix / z _pix ]) (Formula 7)
As a result of the comparison, if the distance already obtained represents a distance closer to the camera, the process for the pixel is terminated as it is, and the process is performed for the next pixel.

On the other hand, as a result of the comparison, if the newly obtained distance represents a distance closer to the camera, the quantization process is performed according to the following equation 8, and the three-dimensional point being processed is projected onto the encoding target camera. The distance information of the position (x _pix / z _pix , y _pix / z _pix ) is updated [step C13].

D ′ _ref [x _pix / z _pix , y _pix / z _pix ] = S _in (z _pix ) (Equation 8)
In the conversion process of FIG. 4, the quantization process is performed in step C13, and the value may be inversely quantized in step C12. In order to reduce the amount of calculation, a distance buffer is separately defined for each pixel position. In step C13, the distance value is directly stored in the distance buffer without quantization, and in step C12, the distance is not subjected to inverse quantization. You may compare using the value of the distance stored in the buffer. In that case, in step C1, the distance value indicating infinity is initialized, and after the comparison in step C15 is not established, the distance value stored in the distance buffer is quantized to generate the converted distance information. To do.

  Even when the reference distance information is converted using the conversion process of FIG. 4, the prediction encoding process performed by the distance information encoding unit 108 stores the reference distance information in the reference distance information memory 107 as described in the first embodiment. Efficient encoding can be performed by encoding while using the stored distance information as a reference.

  When the conversion process of FIG. 4 is used, the distance information stored in the reference distance information memory 107 corresponding to a predetermined index ref is subtracted from the encoding target distance information, and the difference distance information is obtained. May be encoded while predicting from another time already encoded that has been generated in the same manner or difference distance information of another camera. Since the conversion process of FIG. 4 converts not only the distance value but also the position so as to correspond to the encoding target, most of the difference distance information generated by such subtraction for each pixel becomes zero, and the occlusion This is because efficient encoding can be achieved by predicting and encoding only a portion or the like from already encoded information.

  Furthermore, instead of using only one corresponding to a predetermined index ref, a new information constructed by distance information indicating that the pixel position is closest to the camera from the distance information stored in the reference distance information memory 107 is used. There is also a method of generating the distance information, subtracting the constructed distance information from the encoding target distance information, and encoding the difference distance information. In addition, there is also a method of taking an average value or a median value for each pixel to construct new distance information. By using a plurality of reference distance information in this way, the occlusion portion can be reduced, the influence of reference distance information errors due to noise can be reduced, and efficient coding can be achieved.

  Next, a second example (hereinafter referred to as Example 2) will be described. FIG. 5 shows a configuration diagram of a distance information decoding apparatus according to Embodiment 2 of the present invention. As shown in FIG. 5, the distance information decoding apparatus 200 includes a distance information encoded data input unit 201 that inputs encoded data of distance information to be decoded, and an encoded data memory that stores the input encoded data. 202, a camera information input unit 203 for inputting camera parameters and the like of a camera whose distance information to be decoded is a target, and already decoded distance information that the distance information encoded data to be decoded is referring to in predictive encoding A reference distance information input unit 204 for inputting the reference distance information, a reference camera information input unit 205 for inputting camera parameters and the like of the camera for which the reference distance information is targeted, and reference three-dimensional point restoration means. A reference distance information conversion unit 206 that converts the reference distance information, a reference distance information memory 207 that stores the converted reference distance information, and a reference distance information memory 20 And a distance information decoding unit 208 to actually decode the encoded data stored in the encoded data memory 202 with reference to the distance information stored in the.

  FIG. 6 shows a processing flow executed by the distance information decoding apparatus 200 configured as described above. The processing executed by the distance information decoding device 200 will be described in detail according to this processing flow.

  First, encoded data of distance information to be decoded is input from the encoded distance information data input unit 201 and stored in the encoded data memory 202 [step D1].

Next, information such as camera parameters of the camera based on the distance information to be decoded is input from the camera information input unit 203 [step D2]. In the following, the camera internal parameter matrix for the distance information to be decoded is represented as A _out , the rotation matrix as R _out , and the translation vector as t _out . The definition of the camera parameter is the same as in the first embodiment.

In this embodiment, it is assumed that the distance information is represented as a gray scale image. Information necessary for associating the distance from the camera to the subject with the pixel value is also included in the camera information input in step D2. In addition, since the specific example regarding this information was described in Example 1, it abbreviate | omits here. Hereinafter, a function for associating the distance d with the pixel value is represented as S _out (d).

  Next, reference distance information used when predictive encoding distance information to be decoded is prepared [step D3-D8]. This process is performed for each already decoded distance information used as a reference destination. That is, if the index representing the distance information frame that is a reference candidate is ref and the number of reference candidate frames is numRefs, ref is initialized to 0 [step D3], and 1 is added to ref [step D7], ref Until step NumRefs is reached [step D8], the following steps D4-D6 are repeated.

In the process performed for each reference candidate, first, already decoded reference distance information that becomes a reference candidate is input from the reference distance information input unit 204 [step D4]. In the following, the reference distance information of the index ref is represented as D _ref, and the distance information of a specific region is represented by adding information for identifying a position sandwiched between symbols [].

Next, camera information based on D _ref is input from the reference camera information input unit 205 [step D5]. The camera information here includes information necessary for associating the camera parameter and the distance from the camera to the subject with the pixel value, similarly to the camera information for the decoding target camera.

In the following, the camera internal parameter matrix for D _ref is represented by A _ref , the rotation matrix is represented by R _ref , the translation vector is represented by t _ref , and the function for quantizing the distance d from the camera to the subject is represented by S _ref (d).

When the reference distance information and the reference camera information are obtained, the reference distance information conversion unit 206 converts the reference distance information D _ref and stores it in the reference distance information memory 207 [step D6]. The conversion here is the same as that performed in the first embodiment. It is necessary to replace the subscript in in FIG. 3 and FIG. 4 with the subscript out in the corresponding explanation.

  When the conversion of all the reference distance information is finished, the distance information decoding unit 208 refers to the reference distance information stored in the reference distance information memory 207 while referring to the encoded data stored in the encoded data memory 202. Decode [Step D9]. The decoding method here may be any method as long as it is a decoding method for the encoding method used when generating the input encoded data. For example, MPEG-2 and H.264. In the case of encoding in a format compliant with an international standard format for moving image encoding such as H.264 / AVC, MPEG-2 and H.264. A decoding method compliant with H.264 / AVC is used.

  In the second embodiment, all the reference distance information candidates are converted before the encoded data is analyzed. However, it is necessary when the encoded data is sequentially decoded and the fact of reference is found. You may convert only this part. By doing so, it is possible to prevent unnecessary conversion processing from being performed and reduce the amount of calculation.

  As in the case of the first embodiment, when the conversion process of FIG. 4 is used, the distance information decoding unit 208 uses the distance information stored in the reference distance information memory 207 corresponding to a predetermined index ref. There is a method of decoding the decoding target distance information by decoding the difference distance information subtracted from the decoding target distance information from the encoded data and adding the converted distance information corresponding to the index ref to the difference distance information. .

  Also, instead of using only the one corresponding to the predetermined index ref, a new information constructed by distance information indicating that the pixel position is closest to the camera from the distance information stored in the reference distance information memory 207. Of generating distance information, decoding the distance information generated by subtracting the constructed distance information from the decoding target distance information from the encoded data, and decoding the reference distance information by adding the constructed distance information There is also. In addition, there is also a method of taking an average value or a median value for each pixel to construct new distance information.

  The processing described above can be realized by a computer and a software program, and the program can be provided by being recorded on a computer-readable recording medium or can be provided through a network.

  In the above embodiments, the distance information encoding device and the distance information decoding device have been mainly described. However, the distance information encoding method and the distance information decoding method of the present invention are performed according to steps corresponding to the operations of the respective units of these devices. Can be realized.

  As mentioned above, although embodiment of this invention has been described with reference to drawings, the said embodiment is only the illustration of this invention and it is clear that this invention is not limited to the said embodiment. . Accordingly, additions, omissions, substitutions, and other modifications of the components may be made without departing from the spirit and scope of the present invention.

It is a figure which shows the structure of the distance information encoding apparatus of Example 1. FIG. It is a distance information encoding flowchart in Example 1. 10 is a flowchart of reference distance information conversion processing according to the first and second embodiments. 4 is a reference distance information conversion processing flowchart different from FIG. 3. It is a figure which shows the structure of the distance information decoding apparatus of Example 2. FIG. It is a distance information decoding flowchart in Example 2. It is a figure which shows that the distance with respect to the same to-be-photographed object changes with camera positions and directions. It is a figure which shows that the difference of the distance between cameras differs with a to-be-photographed object. It is a figure which shows that the expression range of required distance differs with cameras.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Distance information encoding apparatus 101 Distance information input part 102 Distance information memory 103 Camera information input part 104 Reference distance information input part 105 Reference camera information input part 106 Reference distance information conversion part 107 Reference distance information memory 108 Distance information encoding part 200 Distance information decoding apparatus 201 Distance information encoded data input unit 202 Encoded data memory 203 Camera information input unit 204 Reference distance information input unit 205 Reference camera information input unit 206 Reference distance information conversion unit 207 Reference distance information memory 208 Distance information decoding unit

Claims (18)

In a distance information encoding method for encoding distance information representing a distance from a camera to a subject with respect to an image taken by a camera,
An encoding target camera information setting step for setting parameters of a camera that has photographed a subject in the encoding target distance information, and encoding target camera information for quantizing the distance from the camera to the subject into pixel values;
A reference distance information setting step for setting reference distance information obtained by decoding the already encoded distance information used for prediction of the encoding target distance information;
A reference camera information setting step for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
Using the reference camera information and the reference distance information, a reference three-dimensional point restoration step of calculating the three-dimensional coordinates of the subject indicated by the reference distance information;
A conversion reference distance setting step for obtaining a distance from the camera indicated by the encoding target camera information to the point represented by the three-dimensional coordinate value calculated in the reference three-dimensional point restoration step;
A conversion reference distance information setting step for setting the conversion reference distance information by quantizing the distance obtained in the conversion reference distance setting step by a quantization method indicated by the encoding target camera information;
A distance information encoding step for predictively encoding the encoding target distance information using the converted reference distance information as a reference target;
A distance information encoding method characterized by comprising: In a distance information encoding method for encoding distance information representing a distance from a camera to a subject with respect to an image taken by a camera,
An encoding target camera information setting step for setting parameters of a camera that has photographed a subject in the encoding target distance information, and encoding target camera information for quantizing the distance from the camera to the subject into pixel values;
A reference distance information setting step for setting reference distance information obtained by decoding the already encoded distance information used for prediction of the encoding target distance information;
A reference camera information setting step for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
Using the reference camera information and the reference distance information, a reference three-dimensional point restoration step of calculating the three-dimensional coordinates of the subject indicated by the reference distance information;
A conversion reference distance setting step for obtaining a distance from the camera indicated by the encoding target camera information to the point represented by the three-dimensional coordinate value calculated in the reference three-dimensional point restoration step;
A conversion destination position setting step for obtaining a position where the three-dimensional point calculated in the reference three-dimensional point restoration step is observed on an image captured by the encoding target camera;
A value obtained by quantizing the distance obtained in the conversion reference distance setting step by the quantization method indicated by the encoding target camera information is set as the conversion reference distance information of the position obtained in the conversion destination position setting step. A conversion reference distance information setting step,
A distance information encoding step for predictively encoding the encoding target distance information using the converted reference distance information as a reference target;
A distance information encoding method characterized by comprising: In the distance information encoding method according to claim 2,
A combined distance information setting step for generating combined distance information in the encoding target camera using one or a plurality of converted reference distance information;
A difference distance information setting step for generating difference distance information by subtracting the combined distance information from the distance information to be encoded;
In the distance information encoding step, encoding the difference distance information;
The distance information encoding method characterized by these. In the distance information encoding method according to claim 3,
In the synthetic distance information setting step, distance information indicating that the conversion reference distance information given for each position is closest to the camera, or an average value or median value of the conversion reference distance information given for each position, The combined distance information of the position,
The distance information encoding method characterized by these. In a distance information decoding method for decoding encoded data of distance information representing a distance from a camera to a subject with respect to an image taken by a camera,
A decoding target camera information setting step for setting parameters of a camera that has taken a subject in the decoding target distance information, and decoding target camera information for quantizing the distance from the camera to the subject into pixel values;
A reference distance information setting step for setting the already decoded distance information used for prediction of the decoding target distance information as reference distance information;
A reference camera information setting step for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
Using the reference camera information and the reference distance information, a reference three-dimensional point restoration step of calculating the three-dimensional coordinates of the subject indicated by the reference distance information;
A conversion reference distance setting step for obtaining a distance from the camera indicated by the decoding target camera information to a point represented by the three-dimensional coordinate value calculated in the reference three-dimensional point restoration step;
A conversion reference distance information setting step for setting the conversion reference distance information by quantizing the distance obtained in the conversion reference distance setting step by a quantization method indicated by the decoding target camera information;
A distance information decoding step of decoding the distance information from the encoded data using the converted reference distance information as a reference object;
A distance information decoding method characterized by comprising: In a distance information decoding method for decoding encoded data of distance information representing a distance from a camera to a subject with respect to an image taken by a camera,
A decoding target camera information setting step for setting parameters of a camera that has taken a subject in the decoding target distance information, and decoding target camera information for quantizing the distance from the camera to the subject into pixel values;
A reference distance information setting step for setting the already decoded distance information used for prediction of the decoding target distance information as reference distance information;
A reference camera information setting step for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
Using the reference camera information and the reference distance information, a reference three-dimensional point restoration step of calculating the three-dimensional coordinates of the subject indicated by the reference distance information;
A conversion reference distance setting step for obtaining a distance from the camera indicated by the decoding target camera information to a point represented by the three-dimensional coordinate value calculated in the reference three-dimensional point restoration step;
A conversion destination position setting step for obtaining a position where the three-dimensional point calculated in the reference three-dimensional point restoration step is observed on an image captured by the encoding target camera;
A value obtained by quantizing the distance obtained in the conversion reference distance setting step by the quantization method indicated by the encoding target camera information is set as the conversion reference distance information of the position obtained in the conversion destination position setting step. A conversion reference distance information setting step,
A distance information decoding step of decoding the distance information from the encoded data using the converted reference distance information as a reference object;
A distance information decoding method characterized by comprising: The distance information decoding method according to claim 6,
A combined distance information setting step for generating combined distance information in the encoding target camera using one or a plurality of converted reference distance information;
A difference distance information decoding step for decoding difference distance information obtained by subtracting the combined distance information from decoding target distance information from encoded data;
In the distance information decoding step, decoding the decoding target distance information by adding the composite distance information to the difference distance information;
The distance information decoding method characterized by this. The distance information decoding method according to claim 7,
In the synthetic distance information setting step, distance information indicating that the conversion reference distance information given for each position is closest to the camera, or an average value or median value of the conversion reference distance information given for each position, The combined distance information of the position,
The distance information decoding method characterized by this. In a distance information encoding device that encodes distance information representing a distance from a camera to a subject with respect to an image taken by a camera,
Encoding target camera information setting means for setting parameters of a camera that has photographed the subject in the encoding target distance information, and encoding target camera information for quantizing the distance from the camera to the subject into pixel values;
Reference distance information setting means for setting reference distance information obtained by decoding the already encoded distance information used for prediction of the encoding target distance information;
Reference camera information setting means for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
A reference three-dimensional point restoration means for calculating the three-dimensional coordinates of the subject indicated by the reference distance information using the reference camera information and the reference distance information;
Conversion reference distance setting means for obtaining a distance from the camera indicated by the encoding target camera information to a point represented by the three-dimensional coordinate value calculated by the reference three-dimensional point restoration means;
Conversion reference distance information setting means for setting the conversion reference distance information by quantizing the distance obtained by the conversion reference distance setting means by a quantization method indicated by the encoding target camera information;
Distance information encoding means for predictively encoding the encoding target distance information using the converted reference distance information as a reference target;
A distance information encoding device comprising: In a distance information encoding device that encodes distance information representing a distance from a camera to a subject with respect to an image taken by a camera,
Encoding target camera information setting means for setting parameters of a camera that has photographed the subject in the encoding target distance information, and encoding target camera information for quantizing the distance from the camera to the subject into pixel values;
Reference distance information setting means for setting reference distance information obtained by decoding the already encoded distance information used for prediction of the encoding target distance information;
Reference camera information setting means for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
A reference three-dimensional point restoration means for calculating the three-dimensional coordinates of the subject indicated by the reference distance information using the reference camera information and the reference distance information;
Conversion reference distance setting means for obtaining a distance from the camera indicated by the encoding target camera information to a point represented by the three-dimensional coordinate value calculated by the reference three-dimensional point restoration means;
Conversion destination position setting means for obtaining a position at which the three-dimensional point calculated by the reference three-dimensional point restoration means is observed on an image photographed by the encoding target camera;
A value obtained by quantizing the distance obtained by the conversion reference distance setting means by the quantization method indicated by the encoding target camera information is set as the conversion reference distance information of the position obtained by the conversion destination position setting means. Conversion reference distance information setting means,
Distance information encoding means for predictively encoding the encoding target distance information using the converted reference distance information as a reference target;
A distance information encoding device comprising: In the distance information encoding device according to claim 10,
Combined distance information setting means for generating combined distance information in the encoding target camera using one or a plurality of converted reference distance information;
Difference distance information setting means for generating difference distance information by subtracting the combined distance information from the encoding target distance information,
The distance information encoding means encodes the difference distance information;
A distance information encoding device characterized by the above. In a distance information decoding device that decodes encoded data of distance information representing a distance from a camera to a subject with respect to an image photographed by the camera,
Decoding target camera information setting means for setting parameters of a camera that has photographed the subject in the decoding target distance information, and decoding target camera information for quantizing the distance from the camera to the subject into pixel values;
Reference distance information setting means for setting the already decoded distance information used for prediction of the decoding target distance information as reference distance information;
Reference camera information setting means for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
A reference three-dimensional point restoration means for calculating the three-dimensional coordinates of the subject indicated by the reference distance information using the reference camera information and the reference distance information;
Conversion reference distance setting means for obtaining a distance from the camera indicated by the decoding target camera information to a point represented by the three-dimensional coordinate value calculated by the reference three-dimensional point restoration means;
Conversion reference distance information setting means for setting the conversion reference distance information by quantizing the distance obtained by the conversion reference distance setting means by a quantization method indicated by the decoding target camera information;
Distance information decoding means for decoding the distance information from the encoded data using the converted reference distance information as a reference object;
A distance information decoding apparatus comprising: In a distance information decoding device that decodes encoded data of distance information representing a distance from a camera to a subject with respect to an image photographed by the camera,
Decoding target camera information setting means for setting parameters of a camera that has photographed the subject in the decoding target distance information, and decoding target camera information for quantizing the distance from the camera to the subject into pixel values;
Reference distance information setting means for setting the already decoded distance information used for prediction of the decoding target distance information as reference distance information;
Reference camera information setting means for setting parameters of the camera that has photographed the subject in the reference distance information, and reference camera information for quantizing the distance from the camera to the subject into pixel values;
A reference three-dimensional point restoration means for calculating the three-dimensional coordinates of the subject indicated by the reference distance information using the reference camera information and the reference distance information;
Conversion reference distance setting means for obtaining a distance from the camera indicated by the decoding target camera information to a point represented by the three-dimensional coordinate value calculated by the reference three-dimensional point restoration means;
Conversion destination position setting means for obtaining a position at which the three-dimensional point calculated by the reference three-dimensional point restoration means is observed on an image photographed by the encoding target camera;
A value obtained by quantizing the distance obtained by the conversion reference distance setting means by the quantization method indicated by the encoding target camera information is set as the conversion reference distance information of the position obtained by the conversion destination position setting means. Conversion reference distance information setting means,
Distance information decoding means for decoding the distance information from the encoded data using the converted reference distance information as a reference object;
A distance information decoding apparatus comprising: The distance information decoding device according to claim 13,
Combined distance information setting means for generating combined distance information in the encoding target camera using one or a plurality of converted reference distance information;
Differential distance information decoding means for decoding difference distance information obtained by subtracting the combined distance information from decoding target distance information from encoded data;
The distance information decoding means decodes the decoding target distance information by adding the composite distance information to the difference distance information,
A distance information decoding apparatus characterized by the above.   A distance information encoding program for causing a computer to execute the distance information encoding method according to any one of claims 1 to 4.   The computer-readable recording medium which recorded the distance information encoding program of Claim 15.   A distance information decoding program for causing a computer to execute the distance information decoding method according to any one of claims 5 to 8.   The computer-readable recording medium which recorded the distance information decoding program of Claim 17.

Images