US20120057633A1

US20120057633A1 - Video Classification Systems and Methods

Info

Publication number: US20120057633A1
Application number: US13/225,202
Authority: US
Inventors: Fang Shi; Biao Wang
Original assignee: Intersil Americas LLC
Current assignee: Intersil Americas LLC
Priority date: 2010-09-02
Filing date: 2011-09-02
Publication date: 2012-03-08
Also published as: US20140369417A1; US20120057629A1; US20120057634A1; US8824554B2; US20120057640A1; US9609348B2

Abstract

Video encoder systems and methods are described that employ table-based content classification. One or more tables relate quantization parameters and P-points for a frame of video that typically comprises macroblocks. A deviation representative of a difference between original and decoded versions of a macroblock is determined, the deviation being further representative of a distribution frequency of the value of a distortion for a P-point. The P-point corresponds to a distortion value that is associated with a minimum rate difference between encoding modes for a macroblock. A motion complexity index is updated using a quantization parameter and non-zero coefficients of the encoded frame. An encoding mode for the macroblock can be retrieved from the tables using the motion complexity index to reference mode information maintained in the tables.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

The present application claims priority from PCT/CN2010/076569 (title: “Video Classification Systems and Methods”) which was filed in the Chinese Receiving Office on Sep. 2, 2010, from PCT/CN2010/076564 (title: “Rho-Domain Metrics”) which was filed in the Chinese Receiving Office on Sep. 2, 2010, from PCT/CN2010/076555 (title: “Video Analytics for Security Systems and Methods”) which was filed in the Chinese Receiving Office on Sep. 2, 2010, and from PCT/CN2010/076567 (title: “Systems And Methods for Video Content Analysis) which was filed in the Chinese Receiving Office on Sep. 2, 2010, each of these applications being hereby incorporated herein by reference. The present Application is also related to concurrently filed U.S. Patent non-provisional applications entitled “Rho-Domain Metrics” (attorney docket no. 043497-0393276), “Video Analytics for Security Systems and Methods” (attorney docket no. 043497-0393277) and “Systems And Methods for Video Content Analysis” (attorney docket no. 043497-0393278), which are expressly incorporated by reference herein.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates the relationship of distortion and rate difference between Intra Inter modes for a given quantization parameter.
FIG. 2 is a flowchart illustrating a content classification based mode decision method.
FIG. 3 is a simplified block schematic illustrating a processing system employed in certain embodiments of the invention.

DETAILED DESCRIPTION

Embodiments of the present invention will now be described in detail with reference to the drawings, which are provided as illustrative examples so as to enable those skilled in the art to practice the invention. Notably, the figures and examples below are not meant to limit the scope of the present invention to a single embodiment, but other embodiments are possible by way of interchange of some or all of the described or illustrated elements. Wherever convenient, the same reference numbers will be used throughout the drawings to refer to same or like parts. Where certain elements of these embodiments can be partially or fully implemented using known components, only those portions of such known components that are necessary for an understanding of the disclosed embodiments will be described, and detailed descriptions of other portions of such known components will be omitted so as not to obscure the disclosed embodiments. In the present specification, an embodiment showing a singular component should not be considered limiting; rather, the invention is intended to encompass other embodiments including a plurality of the same component, and vice-versa, unless explicitly stated otherwise herein. Moreover, applicants do not intend for any term in the specification or claims to be ascribed an uncommon or special meaning unless explicitly set forth as such. Further, certain embodiments of the present invention encompass present and future known equivalents to the components referred to herein by way of illustration.
Video standards such as H.264/AVC employ mode decision as an encoding decision process to determine whether a macroblock (“MB”) is encoded as an intra-prediction mode (“Intra Mode”) or an inter-prediction mode (“Inter Mode”). Rate-distortion optimization techniques are commonly applied in various implementations. When encoding a MB, rate-distortion cost is calculated for both Intra Modes and Inter Modes. The minimum cost mode is selected as the final encoding mode. Depending on the video standard, multiple Intra Modes and Inter Modes are applied. For example, in H.264 standard, there are 4 Intra 16×16 Modes and 9 Intra 4×4 Modes for each MB, and skip macroblock, Inter 16×16 Mode, Inter 16×8, 8×16, 8×8, 8×4, 4×8 and 4×4 Modes for each MB. Rate-distortion cost J is defined as
J=D+λ*R, (1)
where distortion D is defined as the difference between reconstructed MB and original MB, where rate R represents the bits used to encode the current MB, and where coefficient λ is a weighting factor. In one example, the sum of absolute differences (SAD) can be used to quantify distortion.

Rate-Distortion Optimization

Rate-distortion optimization (RDO) techniques can provide a balance of encoding quality and compression ratio. An accurate calculation of rate R in equation (1) is computationally costly and generally involves a dual-pass encoding process which requires the use of hardware resources and which introduces additional delays. Research has been conducted to optimize the calculation of R and to provide a fast rate-distortion balanced mode decision algorithm. However, estimation of bit rate R per MB is generally very costly due to tight pipeline architectures employed in hardware embodiments that provide real-time encoding and multiple-channel encoding.
Accordingly, in certain embodiments, distortion D is used to determine the mode decision when R is omitted from equation (1). Mode optimization typically cannot be achieved by using D alone without considering the bit rate perspective of encoding. For example, in the low-complex background cases, Intra Mode's SAD values for background MBs can be smaller than Inter Mode SAD values: therefore, Intra-mode is typically selected for background MBs. However, Intra Mode encoding usually consumes many more bits than Inter Mode encoding and, consequentially, encoding bits may be wasted and background blocky artifacts can be observed.
Certain embodiments employ a comparison of rate cost distortions J_int ra and J_int er. Based on equation (1), a comparison can be taken as equivalent to the comparison of D_{int ra}+λ*(ΔR)
D_{int er}shown in Equation (2), where λ*(ΔR) (denoted hereafter τ) is the rate difference weighting factor between Intra mode and Inter mode.
J _{int er} =D _{int er}
J _{int ra} =D _{int ra}
Experimental results show there is a pseudo tangent relationship between ΔR and distortion for a given quantization parameter (“QP”) as shown in FIG. 1.
FIG. 1 shows the relationship of ΔR and D for a given QP (in FIG. 1, QP=26), and, in FIG. 1, SAD is used as distortion and ΔR=R_{int ra}−R_{int er}. For the purposes of this description, R_{int ra}represents bit numbers used by the intra mode encoder to encode the current microblock, and R_{int er}represents bit numbers used by the inter mode encoder to encode current microblock. A point P is defined as the point at which Diff_R (ΔR) is equal to the zero point on axis X. Points with D values less than P will consume more bits with Intra Mode encoding (ΔR(=R_{int ra}−r_{int er})>0), while points with D values larger than P will consume less bits with Intra Mode, as shown in the drawing. Experimental results show there is a pseudo tangent relationship between ΔR and distortion for a given QP. The location of P-point is a function of QP and video motion complexity, P-points increase along with the increasing of QP and motion complexity. After a P-point is located, deviation r can be estimated and the Intra Mode/Inter Mode decision can be reached quickly and with greater ease, based on the tangent curve and D value distribution frequency.

Rho-Domain Content Classification

Certain embodiments of the invention use Rho-domain (“ρ-domain”) content classification and certain embodiments provide an innovative ρ-domain metric “θ” and employ systems and methods that apply the metric. In some embodiments, the definition of ρ in ρ-domain can be taken to be the number of non-zero coefficients after transform and quantization in a video encoding process. Additionally, the term “NZ” will be used herein to represent ρ, where NZ can be understood as meaning a number of non-zero coefficients after quantization of each macroblock in video standards such as the H.264 video standard. For the purposes of this description, a ρ-domain deviation metric θ may be defined as a recursive weighted ratio between the theoretical NZ_QP curve and the actual NZ_QP curve. Normalized θ typically fluctuates around 1.0. A value of θ smaller than 1.0 can indicate that the actual encoded bit rate is larger than the expectation, implying that a more complicated motion contextual content has been encountered. In contrast, a value of θ larger than 1.0 indicates that the actual encoded bit rate is smaller than the expectation, implying that smoother motion content has been encountered. Therefore, ρ-domain deviation θ can be used as an indicator to classify video content to high motion complexity, medium, medium-low and low motion complexity categories. Based on motion complexity classification, a fast mode decision algorithm can be employed.

Example of a Content Classification Based Mode Decision Algorithm

In the example of FIG. 2, a content classification based mode decision algorithm is illustrated. The algorithm may be embodied in a combination of hardware and software and may be deployed as instructions and data stored in a non-transitory computer readable media. It will be appreciated that the instructions and data may be configured and/or adapted such that execution of the instructions by a processor cause the processor to perform the method described in FIG. 2.
At step 200, offline trained quantization parameter QP and P-point tables QP_P_T_nare built based on p-domain content classifications, while T_n(T_n=1, 2, 3, . . . 51) denotes different motion complexity classifications. If at step 203 it is determined that a current frame belongs to the first 5 frames of a video sequence, then step 204 is performed next; otherwise step 203 is performed next. At step 204, motion complexity index T_nis initiated based on initial QP and complexity information and the P-point can be found from QP_P_T_ntables. Step 206 can then be performed.
If at step 202 it is identified that the current frame does not belong to the first 5 frames of a video sequence then, at step 203, NZ_QP deviation θ is calculated based on the encoded frame NZ and QP information. At step 205, the motion complexity index based on T_nis then recalculated based on deviation θ. A table lookup from QP_P_T_ntables may be performed to find P for the current frame based on weighted previous frame's QP value and content classification index T_nbefore performing step 206.
At step 206, deviation τ is calculated with respect to distortion D based on the tangent relationship of τ and D, the distribution frequency of D, and the location of P-point. A mathematical model φ can be established as a function of P-point, D and QP for each motion complexity class to represent the cost deviation τ for each MB.
One example of a QP_P_T_nis shown in Table 1, here below:

TABLE 1

	QP_P_Tn table

	QP_P_Tn :
	static int MD_P_TABLE[ ][ ]={
	//{T1,T2,T3,P_point_T1,P_point_T2,P_point_T3}

	{0.8,1.1,2,4,6,6},	//QP = 14
	{0.8,1.1,2,4,6,6},	//QP = 15
	{0.8,1.1,2,5,7,7},	//QP = 16
	{0.8,1.1,2,5,7,7},	//QP = 17
	{0.8,1.1,2,6,8,8},	//QP = 18
	{0.8,1.1,2,6,8,8},	//QP = 19
	{0.8,1.1,2,7,9,9},	//QP = 20
	{0.8,1.1,2,8,9,9},	//QP = 20

	....
	}
	//Listed in the table are relative values.
	//From QP and content classification index Tn, and P_point can
	be obtained form the MD_P_TABLE

At step 208, mode decisions for each MB of current frame can be taken. Inter Mode RD cost J_{int er}can be replaced by D as shown in equation (2) and Intra Mode cost J_{int ra}can be replaced by D+τ, where τ is derived from experimental model φ as described at step 206. A winning mode may be selected as the mode which yields the minimum mode cost J_min. The process is typically repeated until it is determined at 210 the encoding of the current frame is finished.
In certain embodiments, the mode-decision algorithm, QP_P_T_ntable and deviation model φ are built offline from experimental results. Motion classification index T_nand its corresponding methods are described in a related, concurrently filed application titled “ρ-domain metrics θ and its applications.” The video classification based mode decision algorithms, systems and methods described herein can provide a very cost efficient, fast and robust alternative approach compared with conventional systems that tend to be computationally costly, and which are usually involve dual-pass encoding mode decision algorithms. In certain embodiments of the present invention. A fast table-lookup method is used to get a P-point value. From the P-point, QP and content classification index T_n, and MB cost deviation τ can be obtained from a selected experimental model φ. Mode decisions can be made efficiently by inserting τ into equation (2).

System Description

Turning now to FIG. 3, certain embodiments of the invention employ a processing system that includes at least one computing system 30 deployed to perform certain of the steps described above. Computing system 30 may be a commercially available system that executes commercially available operating systems such as Microsoft Windows®, UNIX or a variant thereof, Linux, a real time operating system and or a proprietary operating system. The architecture of the computing system may be adapted, configured and/or designed for integration in the processing system, for embedding in one or more of an image capture system, communications device and/or graphics processing systems. In one example, computing system 30 comprises a bus 302 and/or other mechanisms for communicating between processors, whether those processors are integral to the computing system 30 (e.g. 304, 305) or located in different, perhaps physically separated computing systems 300. Typically, processor 304 and/or 305 comprises a CISC or RISC computing processor and/or one or more digital signal processors. In some embodiments, processor 304 and/or 305 may be embodied in a custom device and/or may perform as a configurable sequencer. Device drivers 303 may provide output signals used to control internal and external components and to communicate between processors 304 and 305.
Computing system 30 also typically comprises memory 306 that may include one or more of random access memory (“RAM”), static memory, cache, flash memory and any other suitable type of storage device that can be coupled to bus 302. Memory 306 can be used for storing instructions and data that can cause one or more of processors 304 and 305 to perform a desired process. Main memory 306 may be used for storing transient and/or temporary data such as variables and intermediate information generated and/or used during execution of the instructions by processor 304 or 305. Computing system 30 also typically comprises non-volatile storage such as read only memory (“ROM”) 308, flash memory, memory cards or the like; non-volatile storage may be connected to the bus 302, but may equally be connected using a high-speed universal serial bus (USB), Firewire or other such bus that is coupled to bus 302. Non-volatile storage can be used for storing configuration, and other information, including instructions executed by processors 304 and/or 305. Non-volatile storage may also include mass storage device 310, such as a magnetic disk, optical disk, flash disk that may be directly or indirectly coupled to bus 302 and used for storing instructions to be executed by processors 304 and/or 305, as well as other information.
In some embodiments, computing system 30 may be communicatively coupled to a display system 312, such as an LCD flat panel display, including touch panel displays, electroluminescent display, plasma display, cathode ray tube or other display device that can be configured and adapted to receive and display information to a user of computing system 30. Typically, device drivers 303 can include a display driver, graphics adapter and/or other modules that maintain a digital representation of a display and convert the digital representation to a signal for driving a display system 312. Display system 312 may also include logic and software to generate a display from a signal provided by system 300. In that regard, display 312 may be provided as a remote terminal or in a session on a different computing system 30. An input device 314 is generally provided locally or through a remote system and typically provides for alphanumeric input as well as cursor control 316 input, such as a mouse, a trackball, etc. It will be appreciated that input and output can be provided to a wireless device such as a PDA, a tablet computer or other system suitable equipped to display the images and provide user input.
In certain embodiments, computing system 30 may be embedded in a system that captures and/or processes images, including video images. In one example, computing system may include a video processor or accelerator 317, which may have its own processor, non-transitory storage and input/output interfaces. In another example, video processor or accelerator 317 may be implemented as a combination of hardware and software operated by the one or more processors 304, 305. In another example, computing system 30 functions as a video encoder, although other functions may be performed by computing system 30. In particular, a video encoder that comprises computing system 30 may be embedded in another device such as a camera, a communications device, a mixing panel, a monitor, a computer peripheral, and so on.
According to one embodiment of the invention, portions of the described invention may be performed by computing system 30. Processor 304 executes one or more sequences of instructions. For example, such instructions may be stored in main memory 306, having been received from a computer-readable medium such as storage device 310. Execution of the sequences of instructions contained in main memory 306 causes processor 304 to perform process steps according to certain aspects of the invention. In certain embodiments, functionality may be provided by embedded computing systems that perform specific functions wherein the embedded systems employ a customized combination of hardware and software to perform a set of predefined tasks. Thus, embodiments of the invention are not limited to any specific combination of hardware circuitry and software.
The term “computer-readable medium” is used to define any medium that can store and provide instructions and other data to processor 304 and/or 305, particularly where the instructions are to be executed by processor 304 and/or 305 and/or other peripheral of the processing system. Such medium can include non-volatile storage, volatile storage and transmission media. Non-volatile storage may be embodied on media such as optical or magnetic disks, including DVD, CD-ROM and BluRay. Storage may be provided locally and in physical proximity to processors 304 and 305 or remotely, typically by use of network connection. Non-volatile storage may be removable from computing system 304, as in the example of BluRay, DVD or CD storage or memory cards or sticks that can be easily connected or disconnected from a computer using a standard interface, including USB, etc. Thus, computer-readable media can include floppy disks, flexible disks, hard disks, magnetic tape, any other magnetic medium, CD-ROMs, DVDs, BluRay, any other optical medium, punch cards, paper tape, any other physical medium with patterns of holes, RAM, PROM, EPROM, FLASH/EEPROM, any other memory chip or cartridge, or any other medium from which a computer can read.
Transmission media can be used to connect elements of the processing system and/or components of computing system 30. Such media can include twisted pair wiring, coaxial cables, copper wire and fiber optics. Transmission media can also include wireless media such as radio, acoustic and light waves. In particular radio frequency (RF), fiber optic and infrared (IR) data communications may be used.
Various forms of computer readable media may participate in providing instructions and data for execution by processor 304 and/or 305. For example, the instructions may initially be retrieved from a magnetic disk of a remote computer and transmitted over a network or modem to computing system 30. The instructions may optionally be stored in a different storage or a different part of storage prior to or during execution.
Computing system 30 may include a communication interface 318 that provides two-way data communication over a network 320 that can include a local network 322, a wide area network or some combination of the two. For example, an integrated services digital network (ISDN) may used in combination with a local area network (LAN). In another example, a LAN may include a wireless link. Network link 320 typically provides data communication through one or more networks to other data devices. For example, network link 320 may provide a connection through local network 322 to a host computer 324 or to a wide are network such as the Internet 328. Local network 322 and Internet 328 may both use electrical, electromagnetic or optical signals that carry digital data streams.
Computing system 30 can use one or more networks to send messages and data, including program code and other information. In the Internet example, a server 330 might transmit a requested code for an application program through Internet 328 and may receive in response a downloaded application that provides or augments functional modules such as those described in the examples above. The received code may be executed by processor 304 and/or 305.

Additional Descriptions of Certain Aspects of the Invention

The foregoing descriptions of the invention are intended to be illustrative and not limiting. For example, those skilled in the art will appreciate that the invention can be practiced with various combinations of the functionalities and capabilities described above, and can include fewer or additional components than described above. Certain additional aspects and features of the invention are further set forth below, and can be obtained using the functionalities and components described in more detail above, as will be appreciated by those skilled in the art after being taught by the present disclosure.
Certain embodiments of the invention provide video encoder systems and methods. In some of these embodiments, the encoder systems employ content classification. Some of these embodiments comprise maintaining one or more tables relating quantization parameters and P-points for a frame of video. In some of these embodiments, the frame comprises one or more macroblocks. Some of these embodiments comprise calculating a deviation representative of a difference between original and decoded versions of a macroblock. Some of these embodiments comprise calculating a deviation representative of a distribution frequency of the value of a distortion. Some of these embodiments comprise calculating a deviation representative of the location of a P-point. In some of these embodiments, the P-point corresponds to a distortion value that is associated with a minimum rate difference between encoding modes for a macroblock. Some of these embodiments comprise updating a motion complexity index using a quantization parameter and a number of non-zero coefficients of the encoded frame. Some of these embodiments comprise selecting an encoding mode for the macroblock using the motion complexity index to reference mode information maintained in the one or more tables.
In some of these embodiments, the selected mode yields a least cost encoding. In some of these embodiments. In some of these embodiments, the deviation comprises a weighted difference of estimated distortion and measured distortion for a selected quantization parameter value. In some of these embodiments, the deviation is normalized. In some of these embodiments, calculating the deviation representative of the difference between original and decoded versions of a macroblock is based on a tangential relationship between the distortion and a rate difference between the encoding modes. In some of these embodiments, each P-point corresponds to a distortion value is associated with no rate difference between encoding modes for the macroblock. In some of these embodiments, the motion complexity index is initiated during receipt of an initial number of frames in a video sequence. In some of these embodiments, there are at least 5 frames in the initial number of frames in the video sequence.
Some of these embodiments comprise modeling cost of deviation for each motion complexity class for each macroblock as a function of P-point, distortion and quantization parameter. Some of these embodiments comprise looking up a P-point for a current frame using a weighted quantization parameter value of a previous frame. In some of these embodiments, the encoding modes comprise an inter-prediction mode and an intra-prediction mode. In some of these embodiments, the encoding modes are defined by the H.264 video standard.
Certain embodiments of the invention provide a video encoder 317 (see FIG. 3). Some of these embodiments comprise a plurality of tables relating quantization parameters and encoding modes for a video frame. Some of these embodiments comprise a content classifier that selects an encoding mode for a macroblock of the video frame from the plurality of tables using a deviation representative of difference between original and decoded versions of the macroblock. Some of these embodiments comprise a processor that maintains a motion complexity index using a quantization parameter and non-zero coefficients of the encoded frame. In some of these embodiments, the motion complexity index is operable to select an encoding mode based on the motion complexity of the frame. In some of these embodiments, the selected mode yields a least cost encoding for the frame. In some of these embodiments, the selected mode yields a least cost encoding for the macroblock. In some of these embodiments, each P-point corresponds to a distortion value that is associated with a minimum rate difference between encoding modes for a macroblock.
Although the present invention has been described with reference to specific exemplary embodiments, it will be evident to one of ordinary skill in the art that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the invention. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense.

Claims

What is claimed is:

1. A method of content classification in a video encoder, comprising:

calculating a deviation representative of a difference between original and decoded versions of a macroblock in a frame of video, a distribution frequency of the value of a distortion and the location of a P-point, wherein the macroblock is associated with a bit rate representing bits used to encode the macroblock, and wherein a P-point represents a point in the frame at which a rate of change of bit rate is equal to zero;

updating a motion complexity index using a quantization parameter and a number of non-zero coefficients in the macroblock when encoded; and

selecting an encoding mode for the macroblock using the motion complexity index to reference mode information maintained in one or more tables relating quantization parameters to one or more P-points for the frame of video, wherein the mode is selected to yield a least cost encoding,

wherein the frame comprises a plurality of macroblocks, each macroblock associated with a bit rate representing bits used to encode the each macroblock, and

wherein each P-point corresponds to a distortion value that is associated with a minimum rate difference between encoding modes for a macroblock.

2. The method of claim 1, wherein the deviation comprises a weighted difference of estimated distortion and measured distortion for a selected quantization parameter value.

3. The method of claim 1, wherein the deviation is normalized.

4. The method of claim 1, wherein calculating the deviation representative of the difference between original and decoded versions of a macroblock is based on a tangential relationship between the distortion and a rate difference between the encoding modes.

5. The method of claim 1, wherein each P-point corresponds to a distortion value that is associated with no rate difference between encoding modes for the macroblock.

6. The method of claim 1, wherein the motion complexity index is initiated during receipt of an initial number of frames in a video sequence.

7. The method of claim 6, wherein the initial number of frames in the video sequence comprises 5 frames.

8. The method of claim 1, further comprising modeling a cost of deviation for each motion complexity class for each macroblock as a function of P-point, distortion and quantization parameter.

9. The method of claim 1, further comprising looking up a P-point for a current frame using a weighted quantization parameter value of a previous frame.

10. The method of claim 1, wherein the encoding modes comprise an inter-prediction mode and an intra-prediction mode.

11. The method of claim 1, wherein the encoding modes are defined by the H.264 video standard.

12. A video encoder, comprising:

non-transitory storage adapted to maintain a plurality of tables relating quantization parameters and encoding modes for a video frame; and

a content classifier that selects an encoding mode for a macroblock of the video frame from the plurality of tables using a deviation representative of a difference between original and decoded versions of the macroblock; and

wherein the video encoder maintains a motion complexity index corresponding to a quantization parameter and non-zero coefficients of the encoded frame, the motion complexity index being operable to select the encoding mode as a function of the motion complexity of the video frame, wherein the selected encoding mode yields a least-cost encoding.

13. The video encoder of claim 12, wherein the deviation is represented by a function of a P-point, a distortion and a quantization parameter, wherein each P-point corresponds to a distortion value that is associated with a minimum rate difference between encoding modes for the macroblock.

14. A non-transitory computer-readable medium encoded with data and instructions wherein the data and instructions, when executed by a processor of a video encoder, cause the video encoder to perform a content classification method comprising:

calculating a deviation representative of a difference between original and decoded versions of a macroblock of a frame of video, a distribution frequency of the value of a distortion and the location of a minimum point corresponding to a distortion value associated with a minimum rate difference between possible encoding modes for the macroblock;

updating a motion complexity index using a quantization parameter and a number of non-zero coefficients in the encoded macroblock; and

selecting an encoding mode for the macroblock using the motion complexity index to reference mode information maintained in one or more tables by the video encoder, the one or more tables relating quantization parameters and minimum points for the frame, wherein each macroblock of the frame is associated with a bit rate representing bits used to encode the each macroblock, and wherein each minimum point represents a point in the frame at which a rate of change of bit rate is equal to zero.

15. The non-transitory computer-readable medium of claim 14, wherein the deviation comprises a weighted difference of estimated distortion and measured distortion for a selected quantization parameter value, and wherein the selected mode yields a least cost encoding.

16. The non-transitory computer-readable medium of claim 15, wherein the deviation comprises a weighted difference of estimated distortion and measured distortion for a selected quantization parameter value and wherein calculating the deviation representative of the difference between original and decoded versions of a macroblock includes determining a tangential relationship between the distortion and a rate difference between the encoding modes.

17. The non-transitory computer-readable medium of claim 14, wherein the method further comprises modeling cost of deviation for each motion complexity class for each macroblock as a function of minimum point, distortion and quantization parameter.

18. The non-transitory computer-readable medium of claim 14, wherein the method further comprises looking up a minimum point for a current frame using a weighted quantization parameter value of a previous frame.

19. The non-transitory computer-readable medium of claim 14, wherein the encoding modes comprise an inter-prediction mode and an intra-prediction mode.

20. The non-transitory computer-readable medium of claim 14, wherein the encoding modes are defined by the H.264 video standard.