JP3970267B2 - Moving picture coding apparatus and moving picture coding method - Google Patents

Description

  The present invention relates to video coding, and relates to ITU-T Recommendation H.264. The present invention relates to a moving image coding apparatus and method for performing coding by a method typified by 26x or ISO / IEC standard MPEG. The moving picture coding apparatus and the moving picture coding method according to the present invention relate to a device mounted on, for example, a mobile phone.

  A conventional image encoding method will be described below using MPEG-4 as an example. In general, an image encoding method represented by MPEG-4 compresses data using a spatial / temporal correlation with respect to an input image signal. Then, based on the data obtained by the spatial compression / temporal compression, variable length coding is further performed in a predetermined order to generate a bit stream.

  Here, for the following description, the concept of the MPEG-4 frame will be described. In MPEG-4, the entire image to be displayed (composite image) is composed of images (objects) of a plurality of image sequences. Therefore, the screen of each image sequence at each display time is displayed as a video object plane (hereinafter referred to as a video object plane). , “VOP”), and is distinguished from the frames in MPEG-1 and MPEG-2. When the entire display image is composed of images of one image series, the VOP and the frame match.

  The VOP has a luminance signal and a color difference signal, and is composed of a plurality of macro blocks. The macro block is composed of 16 pixels in each of the vertical and horizontal directions with respect to the luminance signal, and in the image encoding in MPEG-4, the information amount is compressed by a method such as spatial compression or temporal compression in units of the macro block.

  Spatial compression is performed by quantizing a signal after being transformed from the time domain to the frequency domain by a discrete cosine transform (hereinafter referred to as “DCT”), which is a kind of orthogonal transform. Motion compensation is used for dynamic compression.

  In addition, in the data compression method in units of VOPs, time is determined from spatial intra-frame coding (hereinafter referred to as “intra coding”) that is encoded only by spatial compression within the same screen, and correlation between screens. There is inter-screen coding (hereinafter referred to as “inter-coding”) which is coded using dynamic compression.

  In general, an intra-coded VOP is referred to as I (Intra) -VOP (hereinafter referred to as “I-VOP”). In addition, an inter-encoded VOP that is obtained by performing motion compensation using only a VOP previously encoded as a reference VOP is referred to as P (Predictive) -VOP (hereinafter, “P- VOP encoded by performing bi-directional motion compensation using a VOP encoded before and after in time as a reference VOP is referred to as B (Bi-directionally predictive) -VOP (hereinafter referred to as VOP). , “B-VOP”). However, for example, in the case of an MPEG-4 simple profile, only I-VOP and P-VOP are used, and B-VOP is not used. The following description will be mainly given with respect to this MPEG-4 simple profile.

  In a moving image encoding device, a bit stream having a specified code amount must be output according to a predetermined encoding parameter. Further, the VBV buffer occupancy is assumed on the encoding device side so that the buffer on the decoding device side (virtual buffer verifier, hereinafter referred to as “VBV buffer”) that receives the bit stream does not cause overflow or underflow. Thus, the generated code amount must be controlled.

  The generated code amount is controlled by a quantization parameter (quantization step size) used for quantizing the DCT coefficient set for each macroblock constituting the VOP. Therefore, the generated code amount is controlled in units of VOPs. In general, when the quantization parameter is increased, the generated code amount is decreased, and when the quantization parameter is decreased, the generated code amount is increased. That is, the generated code amount and the quantization parameter are in an inversely proportional relationship. It is possible to change the generated code amount using this property.

  In addition, since the amount of generated code is controlled in units of VOPs, it is common to set quantization parameters for each VOP type such as I-VOP and P-VOP and control them individually for the sake of simplicity of control. is there.

  A technique for providing quantization parameters for each VOP type as described above and individually controlling them is described in Patent Document 1, for example.

Japanese Patent Laid-Open No. 2002-262297 (page 5-7, FIG. 1)

  The problem to be solved by the present invention is that when setting quantization parameters for each VOP type and controlling them individually, the image quality may be visually unnatural depending on the image content. is there. In general, in consideration of handling of encoded data in image editing, etc., an I-VOP is set at intervals (cycles) corresponding to a certain number of frames, and from I-VOP to the next I-VOP During this period, encoding control is performed such that encoding is performed using P-VOP. If the I-VOP interval is, for example, 2 seconds, the I-VOP insertion interval is 30 frames at a frame rate of 15 frames / second.

  Since the image quality of I-VOP is generally important, the initial value of the quantization parameter of I-VOP is set to a value smaller than the quantization parameter of P-VOP. That is, since quantization is performed with a small value, a large amount of code is allocated. When the quantization parameter is set in this way, in an image having a certain amount of motion, a large amount of code is required for the P-VOP, and the P-VOP quantization parameter is larger than the I-VOP quantization parameter. The encoding is performed while substantially maintaining the above state.

  However, when an image with little motion, such as an image close to a still image, is encoded, the P-VOP code amount may be reduced, and the P-VOP quantization parameter value may become extremely small. At this time, since the I-VOP is subjected to intra coding, the quantization parameter value does not become abruptly reduced. Therefore, the assumed I-VOP quantization parameter and the P-VOP quantization The magnitude relationship between the parameter setting values may be reversed, or the difference between the I-VOP quantization parameter and the P-VOP quantization parameter may become larger.

As described above, when the I-VOP quantization parameter and the setting value of the P-VOP quantization parameter are reversed and the difference between them increases, when an image close to a still image is encoded, In some cases, the image quality degradation of the image becomes noticeable.
The present invention has been made to solve the above-described problems. Even when an image with little motion, such as an image close to a still image, is encoded, the visual in the VOP (or frame) encoded in the screen is displayed. It is an object of the present invention to provide a moving image encoding apparatus that can perform high-quality encoding as a whole without deterioration in image quality.

The present invention
Coding means for selectively performing intra-frame coding or inter-frame coding for each frame or VOP of the input moving image;
Code amount detection means for obtaining the code amount of each frame or VOP encoded by the encoding means;
Encoding control means for setting a quantization parameter used in encoding in the encoding means;
Quantization parameter average value calculating means for calculating an average value for each frame or VOP of the quantization parameter set by the encoding control means;
In the intra-frame encoding of one frame or VOP, the average value of the quantization parameter set for the frame or VOP that was encoded immediately before the frame and the frame or VOP that was encoded immediately before the inter-frame Coding condition analysis means for determining whether to increase the target code amount based on the average value of the quantization parameter set for
The encoding control means provides a moving picture encoding apparatus, wherein the quantization parameter for the one frame or VOP is set based on the determination in the encoding condition analysis means. .

  According to the present invention, when encoding an image with little motion, such as an image close to a still image, by analyzing an encoding condition and assigning an appropriate code amount based on the analysis, an intra-screen encoding is performed. There is an effect that high-quality encoding can be performed as a whole without any visual image quality degradation in a frame or VOP.

  In the embodiment described below, an input moving image is encoded in units of VOPs, but the present invention can be similarly applied to the case of encoding in units of frames. The following embodiments correspond to the MPEG-4 simple profile, but the present invention is not limited to this.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a moving picture coding apparatus according to Embodiment 1 of the present invention, FIG. 2 is a view showing a configuration of a frame memory of the moving picture coding apparatus according to Embodiment 1, and FIG. FIG. 4 shows a VOP encoding processing procedure of the moving picture encoding apparatus according to the first embodiment, FIG. 4 is a block diagram illustrating an example of the encoding control unit of FIG. 1, and FIG. 5 is a moving picture encoding according to the first embodiment. It is a figure which shows the encoding condition analysis processing procedure in an apparatus.

  As shown in FIG. 1, the moving image encoding apparatus 100 according to the first embodiment receives an image signal transmitted by wired communication or wireless communication as a VOP, and performs an image encoding process on the input VOP. Therefore, the input VOP is divided into macro blocks, encoded, and output as a bit stream.

  As shown in FIG. 1, a moving image encoding apparatus 100 according to the present embodiment includes a frame memory 1, a change amount detection unit 2, a scene change detection unit 3, an encoder 4, a buffer 5, An amount detection unit 6, an encoding control unit 7, an average quantization parameter calculation unit 8, and an encoding condition analysis unit 9 are provided.

  The frame memory 1 temporarily stores VOPs that are sequentially input from the outside. As shown in FIG. 2, the frame memory 1 is capable of storing successive (continuous) VOPs, and includes a first area 1a for storing data of one VOP (i-th VOP), It is composed of two areas including a second area 1b for storing data of the next VOP ((i + 1) th VOP). Each time one VOP is input from the outside, the VOP in the areas 1a and 1b of the frame memory 1 is rewritten, and the latest (current) VOP ((i + 1) th VOP) is stored in the second area 1b. The previous (immediately preceding) VOP is stored in the first area 1a and the state is maintained. In other words, normally (during continuous operation), the frame memory 1 is maintained in a state where data of two adjacent VOPs are stored.

  The change amount detection unit 2 reads the i-th and (i + 1) -th VOPs from the frame memory 1, that is, the VOPs that follow the moving image captured in the frame memory 1, and the feature amount between them, for example, the luminance signal Detect the amount of change. As the amount of change, for example, the absolute value of the difference in the luminance signal of the VOP correlation is obtained for the same pixel, and the integrated value (total) of the absolute value over the entire VOP is obtained.

  The scene change detection unit 3 detects a scene change based on the change amount of the feature amount between the VOPs detected by the change amount detection unit 2 (determining whether a scene change has been detected for each VOP, that is, each A determination is made as to whether or not the VOP is a scene change VOP), and the value of the scene change signal SG is set based on the detection result (or determination result). That is, when a scene change is detected, the scene change signal SG is set to a first value, for example, “1”, and otherwise, the scene change signal SG is set to a second value, for example, “0”. .

The encoder 4 is supplied with the (i + 1) -th VOP (ie, “current” VOP) from the frame memory 1, and the encoder 4 encodes the input VOP in units of macroblocks.
The buffer 5 temporarily stores the bit stream output from the encoder 4.
The code amount detection unit 6 obtains the code amount of each frame encoded by the encoder 4 by counting the generated code amount for each VOP from the bit stream stored in the buffer 5. .

  The encoding control unit 7 sets a target code amount in encoding performed by the encoder 4 and determines a quantization parameter (quantization step size). At this time, the target code amount is set for each VOP type, that is, individually for the I-VOP and the P-VOP, and the quantization parameter is determined.

  The average quantization parameter calculation unit 8 calculates the average value for each VOP of the quantization parameter for each VOP set by the encoding control unit 7.

  The encoding condition analysis unit 9 is based on the value of the sea change signal SG output from the scene change detection unit 3 and the average quantization parameter value for each VOP output from the average quantization parameter calculation unit 8. Coding conditions (addition of code amount accompanying scene change for VOP coded immediately before, IQa average value of quantization parameter for I-VOP coded just before, P-VOP coded just before By analyzing the quantization parameter average value PQa), the coding state is determined, and based on this, the coding control unit 7 performs control so that appropriate code amount allocation is performed. That is, it is determined whether or not the code amount is to be increased, and the value of the additional request signal WS is set based on the determination result. For example, when it is determined that the premium is required, the premium request signal WS is set to a first value, for example, “1”, and when it is determined that the premium is not necessary, the premium request signal WS is set to a second value, for example, “0”. Set.

  “VOP encoded immediately before” refers to the latest VOP of VOPs encoded in the past, and “I-VOP encoded immediately before” refers to I-VOP encoded in the past. The latest VOP among the VOPs is referred to as “P-VOP encoded immediately before”. The latest VOP among the P-VOPs encoded in the past is referred to as “Previous VOP”. , “Previous I-VOP”, “Previous P-VOP”, or simply “Previous VOP”, “Previous I-VOP”, “Previous P-VOP”.

  The encoding control unit 7 sets the target code amount and determines the quantization parameter as described above, the code amount Ac of each VOP obtained by the code amount detection unit 6, and the encoding condition analysis unit. Reference is made to the determination result regarding the necessity of extra code amount in FIG.

FIG. 3 is a diagram illustrating a VOP encoding processing operation procedure of the moving image encoding apparatus 100. The outline of the VOP encoding process will be described below with reference to FIG.
First, a feature amount between the current VOP and the previous VOP input to the moving image coding apparatus 100, for example, a change amount of a luminance signal is detected (step S1).

  Next, the presence / absence of a scene change is determined based on the change amount of the feature amount detected in step S1, and the value of the scene change signal SG is set based on the determination result (step S2). For example, when a scene change is detected, the scene change signal SG is set to a first value, eg, “1”, and otherwise, the scene change signal SG is set to a second value, eg, “0”. .

Next, the value of the scene change signal SG for the VOP encoded immediately before, the average quantization parameter value IQa of I-VOP encoded immediately before, and the average quantum of the P-VOP encoded immediately before Based on the coding parameter value IQa, the coding condition is analyzed, and it is determined whether or not the code amount needs to be increased (step S3).
The average quantization parameter values IQa and PQa of the I-VOP and P-VOP encoded immediately before are stored in the encoding condition analysis unit 9. The value of the scene change signal SG for the VOP encoded immediately before is also stored in the encoding condition analysis unit 9.

  Next, the target code amount of the encoding process is set based on the result of the analysis of the encoding condition in step S3, that is, the determination result of whether or not the code amount needs to be increased (step S4).

  Then, the quantization parameter Qp for each macroblock is determined based on the target code amount set in step S4 (step S5), and the current VOP is actually encoded (step S6).

Next, the VOP encoding processing operation of the moving image encoding apparatus 100 will be described in detail with reference to FIGS. 1 and 3.
First, VOPs input by an upper control system that controls each unit of the moving picture encoding apparatus 100 according to the first embodiment are temporarily stored in the frame memory 1. As described above, the frame memory 1 has the first area 1a for storing the data of the i-th VOP and the second area 1b for storing the data of the (i + 1) -th VOP. During operation), the state in which data of two consecutive VOPs is stored is maintained.

  The change amount detection unit 2 reads the i-th and (i + 1) -th VOPs from the frame memory 1, that is, the VOPs that follow the moving image captured in the frame memory 1, and obtains the change amount of the feature amount between them. (Step S1). The scene change detection unit 3 determines that the current VOP is the VOP of the scene change if the change amount is equal to or greater than the reference value based on the value of the change amount between successive VOPs obtained by the change amount detection unit 2. If it does not satisfy the reference value, it is determined that it is not a scene change VOP (step S2).

On the other hand, the encoder 4 is supplied with the (i + 1) th VOP from the frame memory 1 (that is, the “current” VOP), and the encoder 4 encodes the input VOP in units of macroblocks. To do. This encoding is performed according to the quantization parameter set by the encoding control unit 7.
The encoder 4 outputs the encoded bit stream to the buffer 5, and the bit stream is temporarily stored in the buffer 5.

  The bit stream stored in the buffer 5 is output from the moving picture encoding apparatus 100 by the above-described upper control system. Similarly, the bit stream stored in the buffer 5 is also input to the code amount detection unit 6.

The code amount detection unit 6 obtains the code amount of each frame encoded by the encoder 4 by counting the code generation amount for each VOP from the bit stream accumulated in the buffer 5, and calculates the obtained code Data representing the quantity is output to the encoding control unit 7.
The code amount detection in the code amount detection unit 6 is performed at the end of each macroblock encoding, and the accumulated value of the generated code amount of the VOP currently being encoded is detected as the macroblock encoding progresses. The detection result is output to the encoding control unit 7.

The encoding control unit 7 sets a quantization parameter used in encoding in units of macroblocks performed in the encoder 4. First, the quantization parameter is set for the first macroblock of each VOP as follows.
That is, the accumulated value of the code amount at the end of each VOP output from the code amount detection unit 6 is detected and stored as the code amount of the VOP, and the average quantization output from the average quantization parameter calculation unit 8 The parameter value is stored, and the generated code amount of the previous VOP of the same VOP type, the average quantization parameter value of the previous VOP of the same VOP type, and the determination of whether or not the target code amount needs to be increased by the encoding condition analysis unit 9 The target code amount of the current VOP is determined based on the result, and the quantization parameter of the first macroblock of the current VOP is determined based on the target code amount of the current VOP thus determined. That is, for the I-VOP, the generated code amount of the I-VOP encoded immediately before, the average quantization parameter value of the I-VOP encoded immediately before, and the extra by the encoding condition analysis unit 9 The target code amount of the current I-VOP is determined based on the determination result of the necessity of the current I-VOP, and the quantum of the first macroblock of the current I-VOP is determined based on the target code amount of the current I-VOP thus determined. Define the parameter. For the P-VOP, the generated code amount of the P-VOP encoded immediately before, the average quantization parameter value of the P-VOP encoded immediately before, and the encoding condition analysis unit 9 need to pay extra. The target code amount of the current P-VOP is determined based on the determination result of “No”, and the quantization parameter of the first macroblock of the current P-VOP is determined based on the target code amount of the current P-VOP thus determined. Determine.

  Also, the setting of quantization parameters for the second and subsequent macroblocks following the first macroblock of the VOP is performed as follows. That is, the code amount detection unit 6 calculates the code amount allocated to a macroblock that has not yet been encoded in the current VOP based on the accumulated value of the generated code amount at the end of encoding of each macroblock, and is encoded from this. The quantization parameter of the macroblock is determined. In this way, encoding is advanced while adjusting the quantization parameter for each macroblock using the quantization parameter of the first macroblock of the VOP as an initial value.

  For such determination of the quantization parameter, the encoding control unit 7 calculates the cumulative value of the code amount at the end of each VOP output from the code amount detection unit 6 as shown in FIG. The VOP code amount detection means 7a that is captured at the last time and detected as the VOP code amount VAc, and the VOP code amount VAc output from the VOP code amount detection means 7a are stored separately for each VOP type. A storage unit 7b for storing the average quantization parameter values IQa and PQa of each VOP calculated by the average quantization parameter calculation unit 8 for each VOP type, and a code amount and an average quantum stored in the storage unit 7b A target code amount determination unit 7c that determines a target code amount based on the encoding parameter value and the additional request signal WS from the encoding condition analysis unit 9, and at the end of encoding of each macroblock An assigned code amount calculation unit 7d for calculating a code amount assigned to an unencoded macroblock of the current VOP based on a cumulative value of the generated code amount, and a target code for the first macroblock of each VOP The quantization parameter is determined based on the target code amount determined by the amount determining unit 7c, and for macroblocks other than the head of each VOP, based on the allocated code amount calculated by the allocated code amount calculating unit 7d. And a quantization parameter determining unit 7e for determining the quantization parameter of the macroblock to be encoded.

  The encoding control unit 7 outputs the quantization parameter set in this way to the encoder 4 and the average quantization parameter calculation unit 8.

  The average quantization parameter calculation unit 8 calculates I-VOP and P-VOP from the set values of quantization parameters individually set for each of the I-VOP and P-VOP output from the encoding control unit 7. For each VOP, the average quantization parameter value for each VOP is calculated individually.

The encoding condition analysis unit 9 calculates the average value Qa for each VOP of the quantization parameter individually set for each of the I-VOP and the P-VOP calculated by the average quantization parameter calculation unit 8, The average value IQa of the quantization parameter set for the immediately preceding I-VOP, the average value PQa of the quantization parameter set for the immediately preceding P-VOP, and the scene change set by the scene change detection unit 3 Based on the value of the signal SG, the encoding condition is analyzed to determine whether or not the code amount needs to be increased (step S3).
Details of the encoding condition analysis processing will be described later.

  An additional request signal WS indicating the result of determining whether or not the encoding amount analysis unit 9 needs to increase the code amount is output to the encoding control unit 7 and the generated code amount VAc of the previous I-VOP or the previous P-VOP. Based on the average quantization parameter values IQa and PQa of the previous I-VOP or the previous P-VOP, the target code amount of the current VOP is set (step S4), and the quantization parameter corresponding to the target code amount is obtained. This is output to the encoder 4 and the average quantization parameter calculation unit 8 as a quantization parameter for the first macroblock of the VOP (step S5). For the second and subsequent macroblocks, as described above, at the end of encoding of each macroblock, the code amount detector 6 still encodes the current VOP based on the accumulated value of the generated code amount of the current VOP. A code amount allocated to a macroblock that is not yet calculated is calculated, and a quantization parameter of a macroblock to be encoded is determined.

  The encoder 4 encodes the current VOP based on the set value of the quantization parameter set by the encoding control unit 7 (step S6).

  Hereinafter, a method of analyzing the encoding condition in the encoding condition analyzing unit 9 of the moving image encoding apparatus 100, that is, details of step S3 in FIG. 3 will be described with reference to FIG.

  The encoding condition analysis unit 9 calculates the average quantization parameter value of the immediately preceding I-VOP calculated by the average quantization parameter calculation unit 8 and the average quantum of the P-VOP encoded immediately before. Encoding condition based on the encoding parameter value and the value of the scene change signal SG of the previous VOP indicating whether or not the previous VOP was a scene change VOP (stored in the encoding condition analysis unit 9). To determine whether or not an extra code amount is necessary.

  The encoding condition analysis unit 9 first determines whether the VOP to be encoded is an I-VOP or a P-VOP (step S10). In the case of P-VOP, it is determined that normal code amount allocation is performed based on the scene change signal of the current VOP, that is, when the target code amount is set in step S4, the code amount is not increased (step S4). S16). In the case of a scene change VOP, an additional code amount for scene change is performed.

  If it is determined in step S10 that the encoding is I-VOP, the magnitude relationship between the average quantization parameter value IQa of the previous I-VOP and the average quantization parameter value PQa of the previous P-VOP is compared. Then, it is determined whether or not the average quantization parameter value IQa of the previous I-VOP is sufficiently larger than the average quantization parameter value PQa of the previous P-VOP (step S11). Specifically, the average quantization parameter value IQa of the previous I-VOP is larger than the average quantization parameter value PQa of the previous P-VOP by a certain value (first predetermined value) α or more (IQa−PQa> α). That is, whether or not the average quantization parameter value IQa of the previous I-VOP is larger than the average quantization parameter value PQa of the previous P-VOP and the difference is larger than the first predetermined value α. Do.

  Since the image quality of I-VOP is generally important, the value of the quantization parameter of I-VOP is preferably set to the same level or less as the quantization parameter of P-VOP, but if the above determination result is YES, The above desirable relationship is not established. That is, inversion of the quantization parameter value occurs. Such inversion occurs because the generated code amount of P-VOP is small and the quantization parameter value of P-VOP is small. This is an image with little motion such as an image close to a still image. Is considered to be in a state of encoding.

  When it is determined in step S11 that the average quantization parameter value IQa of the previous I-VOP is larger than the average quantization parameter value PQa of the previous P-VOP by a first predetermined value α or more (IQa−PQa> α) Therefore, it is determined that the image is close to a still image, and it is determined that code amount allocation for still images is performed, that is, when the target code amount is set in step S4, the code amount is increased (step S15). ).

  In the code amount allocation for still images, the target code amount of the I-VOP is set more than usual in order to prevent the deterioration of the image quality of the I-VOP from being visually noticeable with respect to the image quality of the P-VOP. Set to increase. For example, when the target code amount of the normal I-VOP is set to twice that of the P-VOP, the code amount allocation for still images is set to about 5 times that of the P-VOP. As a result, the I-VOP quantization parameter is set to a small value by the encoding control unit 7, the image quality in the I-VOP is improved, and even in an image with little motion such as an image close to a still image, Image quality degradation can be prevented.

In step S11, it is not determined that the average quantization parameter value IQa of the previous I-VOP is larger than the average quantization parameter value PQa of the previous P-VOP by a first predetermined value α or more (IQa−PQa> α). In the case (when the determination result in step S11 is NO), the process proceeds to step S12.
In step S11, it is not determined that the average quantization parameter value IQa of the previous I-VOP is larger than the average quantization parameter value PQa of the previous P-VOP by a first predetermined value α or more (IQa−PQa> α). Even in the case (when the determination result in step S11 is NO), I-VOP is determined by the relationship between the average quantization parameter value IQa of the previous I-VOP and the average quantization parameter value PQa of the previous P-VOP. Degradation of image quality may be noticeable. Even in such a case, the processes in steps S12 to S14 are performed for the purpose of preventing deterioration of the image quality of the I-VOP by increasing the code amount for the I-VOP.

  In step S12, it is determined whether or not the previous VOP was a scene change VOP. Since a large amount of code is generated at the time of a scene change, code amount allocation for increasing the amount of code for scene change and preventing image quality deterioration is performed. When code amount allocation for still images is further performed in the next I-VOP As a result, the code amount increases continuously and locally, image quality degradation occurs in other parts, and the overall image quality cannot be maintained. If the previous VOP is a scene change, it corresponds to a screen change. Therefore, even if the quantization parameter value IQa of the next I-VOP is somewhat large, the image quality degradation in the I-VOP is not noticeable. Therefore, here, when it is determined that the previous VOP is a scene change VOP, it is determined that normal code amount allocation is performed, that is, when the target code amount is determined in step S4, the code amount is not increased. (Step S16).

Next, in step S13, whether or not the average quantization parameter value PQa of the previous P-VOP is encoded as a moving image is determined to be sufficiently small, specifically from a certain value (second predetermined value) β. Is smaller (whether PQa <β) or not. That is, if the average quantization parameter value PQa of the previous P-VOP is equal to or greater than the second predetermined value β (if the determination result in step S12 is NO), it is determined that the image has some degree of motion, and normal When determining the target code amount in step S4, it is determined not to increase the code amount (step S16).
If the average quantization parameter value PQa of the previous P-VOP is sufficiently small in step S13 (if PQa <β, that is, if the determination result in step S12 is YES), the process proceeds to step S14.

  Next, in step S14, the magnitude relationship between the average quantization parameter value PQa of the previous P-VOP and the average quantization parameter value IQa of the previous I-VOP is compared, and the average quantization parameter of the previous P-VOP is compared. It is determined whether or not the value PQa is sufficiently larger than the average quantization parameter value IQa of the previous I-VOP. Specifically, the average quantization parameter value PQa of the previous P-VOP is larger than the average quantization parameter value IQa of the previous I-VOP by a certain value (third predetermined value) γ (PQa−IQa> γ). That is, whether or not the average quantization parameter value PQa of the previous P-VOP is larger than the average quantization parameter value IQa of the previous I-VOP and the difference is larger than a third predetermined value γ. Do. The third predetermined value γ is set to a value that does not cause image quality degradation in the I-VOP if there is a difference greater than that. If the determination result in step S14 is YES, it is determined that normal code amount allocation is performed, that is, when the target code amount is determined in step S4, the code amount is not increased (step S16).

  When the determination result in step S14 is NO, that is, when the average quantization parameter value PQa of the previous P-VOP is larger than the average quantization parameter value IQa of the previous I-VOP, or when the difference is small. When the average quantization parameter value IQa of the VOP is equal to the average quantization parameter value PQa of the previous P-VOP, or the average quantization parameter value IQa of the previous I-VOP is larger than the average quantization parameter value PQa of the previous P-VOP In this case, it is determined that the image quality is deteriorated by the I-VOP, it is determined that the image is close to the still image, and the code amount allocation for the still image is performed, that is, the target code amount is determined in step S4. In this case, it is decided to increase the code amount. In other words, in order to prevent the image quality of the I-VOP from becoming noticeably deteriorated with respect to the image quality of the P-VOP, the target code amount of the I-VOP is set to be larger than that in the normal case. That is, in determining the target code amount in step S4, it is determined to increase the code amount (step S15).

  In this way, by analyzing the relationship between the average quantization parameter value IQa of the previous I-VOP and the average quantization parameter value PQa of the previous P-VOP, the P-VOP quantization is small, and the P-VOP quantization is small. The parameter value becomes extremely small, the magnitude relationship between the assumed I-VOP quantization parameter and the set value of the P-VOP quantization parameter is reversed, and further, the I-VOP quantization parameter and the P- When the difference in the VOP quantization parameter is detected and it is determined that the image quality is deteriorated in the I-VOP, the target code amount of the I-VOP is set to be larger than in the normal case. By setting, the quantization parameter value of the I-VOP is set small, the image quality in the I-VOP is improved, and even in an image with little motion such as an image close to a still image, the I-VOP It is possible to prevent the deterioration of image quality.

  In addition, based on the value of the scene change signal SG of the previous VOP, it is analyzed whether a scene change has been detected in the previous VOP (that is, whether the code amount accompanying the scene change has been increased). Is a scene change VOP, and when the code amount is increased for the scene change, control is performed so that the code amount is not further increased in the next I-VOP. It is possible to prevent the code amount from increasing and the image quality of other portions from deteriorating, and to maintain a stable high image quality as a whole.

Note that the order of the determination steps shown in FIG. 5 can be changed. For example, steps S12 and S13 may be interchanged. Summarizing the criteria for determining whether or not to increase the code amount in steps S11 to S14 is as follows.
(A) When it is determined that the average quantization parameter value IQa of the previous I-VOP is larger than the average quantization parameter value PQa of the previous P-VOP by a first predetermined value α or more (IQa−PQa> α) (YES in step S11), it is determined to increase the code amount in determining the target code amount.
(B) The average quantization parameter value IQa of the previous I-VOP is not larger than the first predetermined value α by the average quantization parameter value PQa of the previous P-VOP (not IQa−PQa> α) (step S11) NO), and when a scene change is detected for the previous VOP (YES in step S12), it is determined not to increase the code amount in determining the target code amount.
(C) The average quantization parameter value IQa of the previous I-VOP is not greater than the first predetermined value α or more than the average quantization parameter value PQa of the previous P-VOP (not IQa−PQa> α) (step S11) If the average quantization parameter value of the previous P-VOP is equal to or greater than the predetermined value β (NO in step S13), it is determined not to increase the code amount in determining the target code amount.
(D) The average quantization parameter value IQa of the previous I-VOP is not greater than the first predetermined value α or more than the average quantization parameter value PQa of the previous P-VOP (not IQa−PQa> α) (step S11) NO) and when the average quantization parameter value PQa of the previous P-VOP is larger than the average quantization parameter value IQa of the previous I-VOP by a third predetermined value γ or more (YES in step S14), the target code amount It is determined that no extra code amount is to be performed.
(E) No scene change is detected for the previous VOP (NO in step S12), the average quantization parameter value of the previous P-VOP is not equal to or greater than the predetermined value β (YES in step S13), and the previous P- If the average quantization parameter value PQa of the VOP is not larger than the third predetermined value γ by the average quantization parameter value IQa of the previous I-VOP (NO in step S14), the code amount is increased when determining the target code amount. Decide to do. If NO in step S14, it is naturally YES in step S11. However, even if YES in step S11, step S14 may not be reached because step S14 is YES in step S12 or NO in step S13.

  In the first embodiment, as an example of code amount allocation for still images, when the target code amount of a normal I-VOP is set to twice the target code amount of P-VOP, P-VOP However, the method of setting the target code amount is not limited to the method of setting the integer code ratio between the target code amount of I-VOP and the target code amount of P-VOP in this way. , I-VOP, P-VOP actual generated code amount, I-VOP, P-VOP average quantization parameter value, setting of target code amount based on combination of encoding information such as scene change information It may be done.

  In the first embodiment, the encoding condition is analyzed by using the average quantization parameter value IQa of the previous I-VOP and the average quantization parameter value PQa of the previous P-VOP. A plurality of generalized I-VOPs and P-VOP average quantization parameter values may be used.

  In the first embodiment, the encoding condition is analyzed by the method shown in FIG. 5. However, the present invention is not limited to this. In general, the I-VOP quantization parameter value IQa and the P-VOP are used. Any quantization parameter value PQa can be used as long as it can detect a case where the image quality degradation in the I-VOP is visually outside the allowable range.

  In the first embodiment, the change amount detection unit 2 detects the change amount based on the luminance signal. However, a color signal may be used instead of the luminance signal.

  Further, the amount of change between successive VOPs is obtained from the sum of all the pixels in the screen of the absolute value of the signal difference for each pixel. However, the present invention is not limited to this. Screen feature values such as a maximum value, a minimum value, and a median value may be used.

  In addition, the amount of change between successive VOPs is calculated for the entire screen. However, the present invention is not limited to this. The screen is divided into several parts, calculated within the divided small areas, and a predetermined value. May be compared.

  All these scene change determination conditions may be combined. For example, a combination of these determination conditions, a method of detecting a scene change when at least one of the determination conditions is satisfied, whether a scene change is determined by a majority decision using the determination result of whether each determination criterion is satisfied A method of determining, a method of determining whether or not a scene change is made by majority by weighting the determination result of whether or not each determination condition is satisfied, and detecting a scene change when a predetermined number or more of the determination conditions are satisfied There are a method, a method of detecting a scene change regardless of the determination result based on other determination conditions when a specific determination condition is satisfied, and these are also included in the present invention.

  In the first embodiment, the scene change detection unit 3 detects the scene change based on the amount of change between the current VOP and the previous VOP, but not only the amount of change but also the encoding state, for example, Further, scene change detection may be performed using a skip signal or the like indicating that encoding is not performed by code amount control, that is, encoding of a certain VOP is skipped.

  Furthermore, in Embodiment 1 described above, only P-VOP is performed as inter-frame coding, but the present invention can be applied to not only P-VOP but also B-VOP. In this case, the encoding condition is analyzed based on the average quantization parameter values of I-VOP, P-VOP, and B-VOP, and the quantization parameter is set for each VOP type.

  Hereinafter, according to the coding condition analysis processing flow of FIG. 5 shown in the first embodiment, the coding condition analysis method in the coding condition analysis unit 9 will be described with respect to the average quantization parameter value IQa of the previous I-VOP, Description will be made by applying specific numerical examples for the average quantization parameter value PQa of the previous P-VOP, the determination values α, β, γ, and the scene change signal SG of the previous VOP. In both the following first example and second example, it is assumed that the VOP to be encoded is an I-VOP, and it is determined in step S10 that the I-VOP is encoded.

In the first example, the average quantization parameter value IQa of the previous I-VOP and the average quantization parameter value PQa of the previous P-VOP are compared in step S11 by comparing the magnitude relationship between the average quantization parameter value PQa of the previous I-VOP. This is a case where it is determined that the quantization parameter value IQa is sufficiently larger than the average quantization parameter value PQa of the previous P-VOP.
Average quantization parameter value IQa of previous I-VOP = 11, average quantization parameter value PQa of previous P-VOP = 3, α = 1, β = 6, γ = 0, and previous VOP is scene change VOP .
In step S11, (average quantization parameter value IQa of previous I-VOP) − (average quantization parameter value PQa of previous P-VOP) = 8 and larger than α = 1, so that an image close to a still image It is determined that the encoding is being performed. In this case, the previous VOP is a scene change VOP, but since the average quantization parameter value IQa of the previous I-VOP is sufficiently larger than the average quantization parameter value PQa of the previous P-VOP, the previous VOP and the I-VOP The code amount is determined to be increased as a difference in image quality occurs (step S15).

The second example is different from the first example in step S11 by comparing the magnitude relationship between the average quantization parameter value IQa of the previous I-VOP and the average quantization parameter value PQa of the previous P-VOP. This is a case where it is not determined that the average quantization parameter value IQa of the previous I-VOP is sufficiently larger than the average quantization parameter value PQa of the previous P-VOP.
It is assumed that the previous I-VOP average quantization parameter value IQa = 5, the previous P-VOP average quantization parameter value PQa = 5, α = 1, β = 6, γ = 0, and the previous VOP is not a scene change VOP.
In step S11, (average quantization parameter value IQa of previous I-VOP) − (average quantization parameter value PQa of previous P-VOP) = 0 and smaller than α = 1, so the process proceeds to step S12.
In step S12, since the previous VOP is not a scene change VOP, the process proceeds to step S13.
In step S13, since the average quantization parameter value PQa of the previous P-VOP is smaller than 5 <6 (= β), it is determined that the average quantization parameter value PQa of the previous P-VOP is sufficiently small, and the process proceeds to step S14.

Next, in step S14, the magnitude relationship between the average quantization parameter value PQa of the previous P-VOP and the average quantization parameter value IQa of the previous I-VOP is compared, and the average quantization parameter of the previous P-VOP is compared. It is determined whether or not the value PQa is larger than the average quantization parameter value IQa of the previous I-VOP, and the difference does not cause image quality degradation in the I-VOP.
In this case, since there is no difference between the average quantization parameter value IQa = 5 of the previous I-VOP and the average quantization parameter value PQa = 5 of the previous P-VOP, image quality degradation in the I-VOP is compared with the P-VOP. Then, it is determined that this is a visually conspicuous case, and it is determined to increase the code amount (step S15).

  It should be noted that portions other than the frame memory 1 and the buffer 5 of the moving picture encoding apparatus according to the above-described embodiment, that is, the change amount detection unit 2, the scene change detection unit 3, the encoder 4, and the code amount detection unit in FIG. 6. The encoding control unit 7, the average quantization parameter calculation unit 8, and the encoding condition analysis unit 9 may be realized by software, that is, by a programmed computer.

It is a block diagram which shows the moving image encoder which shows embodiment of this invention. It is a block diagram of the frame memory of the moving image encoder which shows Embodiment 1 of this invention. It is a flowchart which shows the encoding process procedure of VOP of the moving image encoder which shows Embodiment 1 of this invention. It is a block diagram which shows the detail of the encoding control part 7 of FIG. It is a flowchart which shows the encoding condition analysis processing procedure of the moving image encoder which shows Embodiment 1 of this invention.

Explanation of symbols

  1 frame memory, 2 change amount detection unit, 3 scene change detection unit, 4 encoder, 5 buffer, 6 code amount detection unit, 7 encoding control circuit, 8 average quantization parameter calculation unit, 9 encoding condition analysis unit 100 A moving image encoding device.

Claims (10)

Coding means for selectively performing intra-frame coding or inter-frame coding for each frame or VOP of the input moving image;
Code amount detection means for obtaining the code amount of each frame or VOP encoded by the encoding means;
Encoding control means for setting a quantization parameter used in encoding in the encoding means;
Quantization parameter average value calculating means for calculating an average value for each frame or VOP of the quantization parameter set by the encoding control means;
In the intra-frame encoding of one frame or VOP, the average value of the quantization parameter set for the frame or VOP that was encoded immediately before the frame and the frame or VOP that was encoded immediately before the inter-frame Coding condition analysis means for determining whether to increase the target code amount based on the average value of the quantization parameter set for
The encoding control means sets the quantization parameter for the one frame or VOP based on the determination in the encoding condition analysis means ,
The encoding condition analysis means includes
When it is determined that the average quantization parameter value of the frame or VOP that was intra-encoded just before is larger than the average quantization parameter value of the frame or VOP that was inter-encoded just before the first predetermined value Is
A moving picture encoding apparatus, wherein a code amount is increased when determining the target code amount . Further we have a means for detecting the scene change for the encoded frame or VOP,
The encoding condition analysis means includes
The average quantization parameter value of the frame or VOP that was intra-encoded just before is not greater than the first predetermined value than the average quantization parameter value of the frame or VOP that was inter-encoded just before, and ,
When a scene change is detected for the frame or VOP encoded immediately before,
2. The moving picture coding apparatus according to claim 1 , wherein an extra code amount is not performed in determining the target code amount . The encoding condition analysis means includes
The average quantization parameter value of the frame or VOP that was intra-encoded just before is not greater than the first predetermined value than the average quantization parameter value of the frame or VOP that was inter-encoded immediately before, and ,
When the average quantization parameter value of the frame or VOP that has been inter-frame encoded immediately before is equal to or greater than a second predetermined value,
2. The moving picture encoding apparatus according to claim 1, wherein when determining the target code amount, no additional code amount is performed. The encoding condition analysis means includes
The average quantization parameter value of the frame or VOP that was intra-encoded just before is not greater than the first predetermined value than the average quantization parameter value of the frame or VOP that was inter-encoded just before, and ,
When the average quantization parameter value of the frame or VOP that has been inter-frame encoded immediately before is larger than the average quantization parameter value of the frame or VOP that has been intra-frame encoded immediately before by a third predetermined value,
2. The moving picture encoding apparatus according to claim 1, wherein when determining the target code amount, it is determined not to increase the code amount. The encoding condition analysis means includes
No scene change is detected for the frame or VOP encoded immediately before, and
The average quantization parameter value of the frame or VOP that has been inter-coded immediately before is not equal to or greater than a second predetermined value, and
When the average quantization parameter value of the frame or VOP encoded immediately before the screen is not greater than the third predetermined value than the average quantization parameter value of the frame or VOP encoded immediately before the screen,
3. The moving picture coding apparatus according to claim 2 , wherein the code amount is increased when the target code amount is determined. A coding step for selectively performing intra-frame coding or inter-frame coding for each frame or VOP of the input moving image;
A code amount detection step for obtaining a code amount of each frame or VOP encoded in the encoding step;
An encoding control step for setting a quantization parameter used in encoding in the encoding step;
A quantization parameter average value calculating step for calculating an average value for each frame or VOP of the quantization parameter set in the encoding control step;
In the intra-frame encoding of one frame or VOP, the average value of the quantization parameter set for the frame or VOP that was encoded immediately before the frame and the frame or VOP that was encoded immediately before the inter-frame An encoding condition analysis step for determining whether or not to increase the target code amount based on the average value of the quantization parameter set for
The encoding control step sets the quantization parameter for the one frame or VOP based on the determination in the encoding condition analysis step ,
The encoding condition analysis step includes:
When it is determined that the average quantization parameter value of the frame or VOP that was intra-encoded just before is larger than the average quantization parameter value of the frame or VOP that was inter-encoded just before the first predetermined value Is
A moving image coding method, wherein a code amount is increased when determining the target code amount . Further have a step of detecting a scene change for the encoded frame or VOP,
The encoding condition analysis step includes:
The average quantization parameter value of the frame or VOP that was intra-encoded just before is not greater than the first predetermined value than the average quantization parameter value of the frame or VOP that was inter-encoded just before, and ,
When a scene change is detected for the frame or VOP encoded immediately before,
7. The moving picture encoding method according to claim 6 , wherein an extra code amount is not performed in determining the target code amount . The encoding condition analysis step includes:
The average quantization parameter value of the frame or VOP that was intra-encoded just before is not greater than the first predetermined value than the average quantization parameter value of the frame or VOP that was inter-encoded immediately before, and ,
When the average quantization parameter value of the frame or VOP that has been inter-frame encoded immediately before is equal to or greater than a second predetermined value,
7. The moving picture coding method according to claim 6 , wherein the code amount is not increased when determining the target code amount. The encoding condition analysis step includes:
The average quantization parameter value of the frame or VOP that was intra-encoded just before is not greater than the first predetermined value than the average quantization parameter value of the frame or VOP that was inter-encoded just before, and ,
When the average quantization parameter value of the frame or VOP that has been inter-frame encoded immediately before is larger than the average quantization parameter value of the frame or VOP that has been intra-frame encoded immediately before by a third predetermined value,
7. The moving picture encoding method according to claim 6 , wherein it is determined not to increase the code amount when determining the target code amount. The encoding condition analysis step includes:
No scene change is detected for the frame or VOP encoded immediately before, and
The average quantization parameter value of the frame or VOP that has been inter-coded immediately before is not equal to or greater than a second predetermined value, and
When the average quantization parameter value of the frame or VOP encoded immediately before the screen is not greater than the third predetermined value than the average quantization parameter value of the frame or VOP encoded immediately before the screen,
8. The moving picture encoding method according to claim 7 , wherein a code amount is increased when determining the target code amount.

Images