Image Encoding Apparatus and Weighted Prediction Image Encoding Method

Abstract

An embodiment of the present application provides a weighted prediction image coding method, which includes the following operations: calculating a first eigenvalue of a first picture frame, a second eigenvalue of a second picture frame, and a third eigenvalue of a third picture frame; calculating a first offset value according to the first eigenvalue and the second eigenvalue, and calculating a second offset value according to the second eigenvalue and the third eigenvalue; and when both the first offset value and the second offset value are greater than or both are less than a predetermined value, coding the third picture frame using a weighted prediction operation, where the first picture frame, the second picture frame, and the third picture frame are a plurality of picture frames in temporal sequence.

Description

Technical Field

[0001] This application relates to an image encoding device, and more particularly to an image encoding device and a weighted prediction image encoding method that can use simple operations to determine whether to enable weighted prediction operations. Background Art

[0002] Weighted prediction operations can be used to process multiple frames in a video where the brightness of the scene is continuously getting brighter or darker. Existing image encoding techniques need to traverse all pixels of each of the current frame and the reference frame multiple times to generate some parameters for decision-making, and use these parameters to correct the frame, so as to determine whether to use weighted prediction operations to encode the current frame. In this way, the overall computational complexity will be too high, and dependencies (or couplings) of data (such as the aforementioned parameters) will be generated during the computational process, making it difficult to implement this image encoding technique using pipelined hardware. Summary of the Invention

[0003] In some embodiments of this application, one of the objectives of this application is to provide an image encoding device and a weighted prediction image encoding method that can use simple operations to determine whether to enable weighted prediction operations, so as to improve the deficiencies of the prior art.

[0004] In some embodiments of this application, the weighted prediction image encoding method includes the following operations: calculating a first eigenvalue of a first frame, a second eigenvalue of a second frame, and a third eigenvalue of a third frame; calculating a first offset value according to the first eigenvalue and the second eigenvalue, and calculating a second offset value according to the second eigenvalue and the third eigenvalue; and when both the first offset value and the second offset value are greater than or both are less than a predetermined value, encoding the third frame using a weighted prediction operation, where the first frame, the second frame, and the third frame are multiple frames in temporal sequence.

[0005] In some embodiments, the image encoding device includes a preprocessing circuit and a weighted prediction circuit. The preprocessing circuit calculates a first eigenvalue of a first frame, a second eigenvalue of a second frame, and a third eigenvalue of a third frame. The weighted prediction circuit calculates a first offset value according to the first eigenvalue and the second eigenvalue, and calculates a second offset value according to the second eigenvalue and the third eigenvalue, and when both the first offset value and the second offset value are greater than or both are less than a predetermined value, controls an encoding circuit to encode the third frame using a weighted prediction operation, where the first frame, the second frame, and the third frame are multiple frames in temporal sequence.

[0006] The features, implementations, and effects of this application will be described in detail in the following preferred embodiments with reference to the drawings. Description of the Drawings

[0007] Figure 1 is a schematic diagram of an image coding device drawn according to some embodiments of the present application;

[0008] Figure 2 is a flowchart of a weighted prediction image coding method drawn according to some embodiments of the present application;

[0009] Figure 3 is a schematic diagram of an image coding device drawn according to some embodiments of the present application; and

[0010] Figure 4 is a schematic diagram of an image coding device drawn according to some embodiments of the present application. Detailed Description of the Embodiments

[0011] All terms used herein have their ordinary meanings. The definitions of the above terms in commonly used dictionaries are only examples of the use of any of the terms discussed herein, and should not limit the scope and meaning of the present application. Similarly, the present application is not limited only to the various embodiments shown in this specification.

[0012] Regarding the use of "coupled" or "connected" herein, it can refer to two or more components making direct physical or electrical contact with each other, or making indirect physical or electrical contact with each other, and can also refer to two or more components operating or acting on each other. As used herein, the term "circuit" can be a device that processes signals by connecting at least one transistor and / or at least one active or passive component in a certain manner.

[0013] Figure 1 FIG. is a schematic diagram of an image coding device 100 drawn according to some embodiments of the present application. In some embodiments, the image coding device 100 can determine whether to enable weighted prediction operations based on the changes of a series of consecutive frames to be coded in a video. The algorithm of the weighted prediction operations can refer to the detailed calculation methods in the related art, and hereinafter, emphasis will be placed on how to determine whether to enable the weighted prediction operations.

[0014] ​​​​The image encoding device 100 includes a preprocessing circuit 110, a weighted prediction circuit 120, and an encoding circuit 130. The preprocessing circuit 110 can sequentially process the pixel luminance values of multiple frames F1 to Fn in the video VI to obtain luminance information corresponding to each of the frames. For example, the multiple frames F1 to Fn are multiple frames in temporal sequence. In other words, if frame F1 is the first frame, frame F2 can be the second frame that appears after frame F1, and frame F3 can be the third frame that appears after frame F2. And so on, the sequential relationship of these frames F1 to Fn in time should be understood. The preprocessing circuit 110 can sequentially process the pixel luminance values of multiple frames F1 to Fn to obtain multiple luminance information respectively corresponding to the multiple frames F1 to Fn. For example, the preprocessing circuit 110 can process the pixel luminance values of frame F1 to obtain the first luminance information of frame F1, process the pixel luminance values of frame F2 to obtain the second luminance information of frame F2, and process the pixel luminance values of frame F3 to obtain the third luminance information of frame F3. And so on, the related operations of the preprocessing circuit 110 should be understood. The preprocessing circuit 110 can transfer the multiple frames F1 to Fn to the memory 100A, so that the encoding circuit 130 can read the multiple frames F1 to Fn from the memory 100A for encoding. In some embodiments, the memory 100A can be, but is not limited to, a dynamic random access memory.

[0015] The preprocessing circuit 110 can generate an eigenvalue corresponding to a specific frame according to the above-mentioned luminance information. The downsampling circuit 112 can downsample a corresponding frame among multiple frames F1 to Fn to obtain a downsampled frame of the corresponding frame, and obtain the luminance values of the pixels of the downsampled frame as the luminance information of the corresponding frame. For example, the preprocessing circuit 110 may include a downsampling circuit 112 and a feature calculation circuit 114. The downsampling circuit 112 can downsample a corresponding frame among multiple frames F1 to Fn to obtain a downsampled frame of the corresponding frame and obtain the luminance values of at least some of the pixels of the downsampled frame as the luminance information of the corresponding frame. For example, the downsampling circuit 112 can reduce the resolution of frame F1 to obtain a downsampled frame of frame F1, and obtain the luminance values of all pixels from the downsampled frame, thereby determining the first luminance information of frame F1. In other words, the number of pixels in the downsampled frame is less than the number of all pixels in frame F1. For example, the downsampling circuit 112 can perform 1 / 4 downsampling in the horizontal and vertical directions in frame F1 (i.e., each pixel in the downsampled frame corresponds to four pixels in frame F1) to generate a downsampled frame of frame F1, and obtain the luminance values of all pixels in the downsampled frame, where the luminance value of each pixel in the downsampled frame can be, but is not limited to, an average value of the luminance values of the four corresponding pixels in frame F1. By analogy, the downsampling circuit 112 can obtain the second luminance information corresponding to frame F2 and the third luminance information corresponding to frame F3. The above numerical values of the sampling rate for downsampling are only for illustration, and the present application is not limited thereto.

[0016] The feature calculation circuit 114 can generate an eigenvalue corresponding to the specific frame according to the luminance values of at least some of the pixels in the downsampled frame corresponding to the specific frame. In some embodiments, the feature calculation circuit 114 can perform statistical operations according to the luminance information of the downsampled frame (such as the luminance values of at least some of the pixels in the downsampled frame) to generate an eigenvalue of the corresponding frame. For example, the feature calculation circuit 114 can determine the luminance information of the downsampled frame according to the luminance values of at least some of the pixels in the downsampled frame, and perform statistical operations accordingly to determine the eigenvalue of the corresponding frame.

[0017] Specifically, in one example, the feature calculation circuit 114 may calculate the average luminance value of all pixels (or at least some of the pixels) in the downsampled frame corresponding to the frame F1 (for example, the sum of the luminance values of the aforementioned all pixels may be divided by the number of these pixels to determine the average luminance value) to generate the feature value of the frame F1 (labeled as VF1). Similarly, the feature calculation circuit 114 may calculate the average luminance value of all pixels (or at least some of the pixels) in the downsampled frame corresponding to the frame F2 to generate the feature value of the frame F2 (labeled as VF2). And so on, the feature calculation circuit 114 may generate the feature value of the frame F3 (labeled as VF3). The above statistical operation method of generating feature values in an average manner is only for illustration, and this application is not limited thereto. In different embodiments, various statistical operation methods that can be used to process luminance information to generate statistical operation methods sufficient to identify the luminance change of the frame can be used to generate the feature value corresponding to the specific frame. In some embodiments, the feature calculation circuit 114 may be implemented by, but not limited to, digital logic circuits that can perform the above operations.

[0018] The weighted prediction circuit 120 may determine a plurality of offset values OS1 and OS2 according to the foregoing plurality of feature values VF1 to VF3, and determine whether to control the encoding circuit 130 to use weighted prediction operation to encode the current frame according to these offset values corresponding to a plurality of consecutive frames. For example, the weighted prediction circuit 120 may determine the offset value OS1 according to the difference between the feature value VF1 and the feature value VF2, and determine the offset value VF2 according to the difference between the feature value VF2 and the feature value VF3, where the offset value OS1 may be expressed as VF2 - VF1 and the offset value OS2 may be expressed as VF3 - VF2. The weighted prediction circuit 120 may determine whether to enable the weighted prediction operation of the encoding circuit 130 according to the comparison results of the offset value OS1 and a predetermined value (for example, but not limited to, 0) and the offset value OS2 and the predetermined value.

[0019] For example, if both the offset value OS1 and the offset value OS2 are greater than or both are less than the predetermined value, it represents that the brightness of the image content corresponding to the picture frames F1, F2, and F3 is continuously getting brighter or continuously getting darker. Under this condition, the weighted prediction circuit 120 can control the encoding circuit 130 to enable weighted prediction operation to encode the current picture frame (for example, picture frame F3). In this way, more efficient encoding can be provided for picture frames with a gradually changing brightness scene. Or, if the offset value OS1 and the offset value OS2 are not both greater than or not both less than the predetermined value, it represents that the brightness of the image content corresponding to the picture frames F1, F2, and F3 may not change (for example, multiple offset values OS1 and OS2 are all equal to the predetermined value) or there is an obvious bright-dark switch (for example, the offset value OS1 is greater than the predetermined value and the offset value OS2 is less than the predetermined value). Under this condition, the weighted prediction circuit 120 can control the encoding circuit 130 not to enable the weighted prediction operation, but to use the general encoding method to encode the current picture frame (for example, picture frame F3).

[0020] Specifically, the weighted prediction circuit 120 includes a buffer 122, an offset calculation circuit 124, and a decision circuit 126. The buffer 122 can be used to store a plurality of feature values VF1 to VF3 and a plurality of offset values OS1 and OS2. The offset calculation circuit 124 can read out the plurality of feature values VF1 to VF3 from the buffer 122, perform the aforementioned difference operation to obtain the plurality of offset values OS1 and OS2, and store the plurality of offset values OS1 and OS2 into the buffer 122. The decision circuit 126 can compare the offset value OS1 and the offset value OS2 with the predetermined value respectively to determine whether to issue a control signal VC to control the encoding circuit 130 to use the weighted prediction operation to encode the current picture frame. In some embodiments, each of the offset calculation circuit 124 and the decision circuit 126 can be implemented by, but not limited to, digital logic circuits capable of performing the above operations.

[0021] The encoding circuit 130 can determine whether to use weighted prediction operation to encode the picture frame read from the memory 100A according to the control signal VC. For example, when the control signal VC has a specific logic value, the encoding circuit 130 can use the weighted prediction operation to encode the currently read picture frame. Alternatively, when the control signal VC does not have the specific logic value, the encoding circuit 130 can use the original image encoding algorithm to encode the currently read picture frame. In some embodiments, the weighted prediction operation is an algorithm for encoding a picture frame using parameters such as a reference picture frame (e.g., a previous picture frame), a prediction weight, and an offset value. The offset value can be the aforementioned offset value OS1 and offset value OS2, and the weight can be a preset value (e.g., can be, but is not limited to, 64). In some embodiments, the value of the weight can be set considering the implementation complexity of the digital circuit. For example, the value of the weight can be set to a power of 2 so that the digital circuit can perform the weighted prediction operation in a simple manner such as a shift operation.

[0022] Figure 2 The flowchart of a weighted prediction image encoding method 200 is drawn according to some embodiments of the present application. In some embodiments, the weighted prediction image encoding method 200 can be executed by, but is not limited to, Figure 1 the image encoding device 100.

[0023] In operation S210, a first eigenvalue of a first picture frame, a second eigenvalue of a second picture frame, and a third eigenvalue of a third picture frame are calculated, where the first picture frame, the second picture frame, and the third picture frame are a plurality of picture frames in temporal sequence. For example, as described above, the preprocessing circuit 110 can downsample a corresponding picture frame among the picture frames F1 to F3 to obtain a downsampled picture frame, determine the luminance information of the corresponding picture frame according to the luminance values of at least some of the pixels in the downsampled picture frame, and generate a corresponding eigenvalue among the eigenvalue VF1 of the picture frame F1, the eigenvalue VF2 of the picture frame F2, and the eigenvalue VF3 of the picture frame F3 according to the luminance information of the corresponding picture frame.

[0024] In operation S220, a first offset value is calculated according to the first eigenvalue and the second eigenvalue, and a second offset value is calculated according to the second eigenvalue and the third eigenvalue. For example, as described above, the weighted prediction circuit 120 can determine the offset value OS1 according to the difference between the eigenvalue VF1 and the eigenvalue VF2, and can determine the offset value OS2 according to the difference between the eigenvalue VF2 and the eigenvalue VF3.

[0025] In operation S230, it is confirmed whether both the first offset value and the second offset value are greater than or less than a predetermined value. If both the first offset value and the second offset value are greater than or less than the predetermined value, operation S240 is performed. Alternatively, if the first offset value and the second offset value are not both greater than or not both less than the predetermined value, operation S250 is performed. In operation S240, the third picture frame is encoded using weighted prediction arithmetic. In operation S250, the third picture frame is not encoded using weighted prediction arithmetic.

[0026] For example, as described above, when both the offset values OS1 and OS2 are greater than or less than the predetermined value, it represents that the luminance of the image content in the most recent picture frame is gradually increasing or decreasing. Under this condition, the weighted prediction circuit 120 can determine to output the control signal VC to control the encoding circuit 130 to encode the current picture frame (for example, picture frame F3) using weighted prediction arithmetic. Alternatively, when the offset values OS1 and OS2 are not both greater than or not both less than the predetermined value, it represents that the luminance of the image content in the most recent picture frame has no change or undergoes an obvious change. Under this condition, the weighted prediction circuit 120 may not control the encoding circuit 130 to encode the current picture frame using weighted prediction arithmetic, such that the encoding circuit 130 can use the original image encoding arithmetic to encode the current picture frame (for example, picture frame F3).

[0027] The operations of the above weighted prediction image encoding method 200 are only examples and are not limited to being executed in the order of this example. Without departing from the operation modes and scopes of the embodiments of the present application, various operations in the weighted prediction image encoding method 200 can be appropriately added, replaced, omitted, or executed in a different order (for example, they can be executed simultaneously or partially simultaneously).

[0028] In some related technologies, the decision-making process for determining whether to perform weighted prediction operations on an image includes the following operations: (1) traversing the frame to be encoded (i.e., obtaining the data values of all pixels in the frame pixel by pixel, such as including luminance values (Y values), chrominance values (U values), and concentration values (V values)) to sequentially calculate the means and relative differences corresponding to the luminance value, the chrominance value, and the concentration value respectively; (2) calculating weights and offset values based on the means and relative differences corresponding to the luminance value, the chrominance value, and the concentration value respectively; (3) traversing the frame to be encoded and a reference frame to calculate a plurality of sums of absolute differences pixel by pixel, and summing up these sums of absolute differences to obtain a first total; (4) traversing the frame to be encoded and the reference frame again, calculating a plurality of sums of absolute differences pixel by pixel, and using the weights and offset values to correct these sums of absolute differences, and summing up the corrected sums of absolute differences to obtain a second total; and (5) determining whether to use weighted prediction operations to encode the frame to be encoded based on the first total and the second total. In the above technologies, it is necessary to traverse all pixels in the frame multiple times, and calculations need to be performed for luminance values, chrominance values, and concentration values, making the computational complexity quite high. In addition, there is data coupling in the above operations (1), (3), and (4) (i.e., some operations need to use the operation results generated by previous operations; for example, operation (4) needs to use the weights and offset values in operation (2)). Thus, the practices of related technologies are not easily implemented in a pipelined hardware environment.

[0029] Compared with the above technologies, some embodiments of the present application only perform operations on luminance values, and perform downsampling on the frame (i.e., not taking the luminance values of all pixels) and use a simple operation method, thereby greatly reducing the computational complexity and eliminating data coupling. Thus, a weighted prediction image coding method 200 that can be easily implemented in pipelined hardware can be achieved without reducing the accuracy of determining whether to use weighted prediction operations.

[0030] Figure 3 FIG. is a schematic diagram of an image coding apparatus 300 drawn according to some embodiments of the present application. In some embodiments, Figure 1 The preprocessing circuit 110 can be implemented by existing circuits in the system. For example, the preprocessing circuit 110 can be implemented by a part of the circuits in an image signal processor (ISP) in the system and the central processor 300A in the system together. For example, different from Figure 1 The image coding apparatus 100, in this example, the foregoing multiple eigenvalue VF1-VF3 are executed by the central processor 300A in the system, and some functions in the preprocessing circuit 110 can be implemented by a part of the circuits in the image signal processor.

[0031] Specifically, in Figure 3 , the preprocessing circuit 110 includes a sampling circuit 312, an accumulation circuit 314, and a counter 316. The above three circuits can be noise reduction modules in an image signal processor (i.e., the aforementioned partial circuits). The sampling circuit 312 downsamples multiple image frames F1 to Fn to obtain multiple downsampled image frames corresponding to the multiple image frames F1 to Fn respectively, and reads at least a part of the pixel luminance values in each downsampled image frame. The operation of the sampling circuit 312 is similar to Figure 1 's downsampling circuit 112, so it will not be repeated here. On the other hand, in the noise reduction function, the sampling circuit 312 can also traverse or downsample multiple image frames F1 to Fn to obtain relevant information about the noise. In other words, in some embodiments, Figure 1 's operation of the downsampling circuit 112 can be integrated into the sampling circuit used by the existing noise reduction module in the system.

[0032] The accumulation circuit 314 is used to accumulate the luminance values of at least a part of the pixels in a downsampled image frame to generate the total luminance value of the downsampled image frame (equivalent to the luminance information of the downsampled image frame). The counter 316 can count when the sampling circuit 312 downsamples multiple image frames F1 to Fn to count the number of pixels in a downsampled image frame corresponding to each image frame. The central processing unit 300A can obtain relevant information such as the total luminance value and the number of pixels, and use this information to generate the aforementioned characteristic values VF1, VF2, and VF3. For example, the accumulation circuit 314 can calculate the total luminance value of the downsampled image frame corresponding to the image frame F1, and the count value of the counter 316 can indicate the number of the pixels in the downsampled image frame corresponding to the image frame F1. In this way, the central processing unit 300A can use the above information to determine the characteristic value VF1 (for example, using the aforementioned average operation). The central processing unit 300A can transmit multiple characteristic values VF1, VF2, and VF3 to the weighted prediction circuit 120 for subsequent operations. For example, the central processing unit 300A can directly store multiple characteristic values VF1, VF2, and VF3 in the buffer 122, or can provide multiple characteristic values VF1, VF2, and VF3 to the offset calculation circuit 124, and store the characteristic values VF1, VF2, and VF3 in the buffer 122 via the offset calculation circuit 124.

[0033] Figure 4 FIG. 400 is a schematic diagram of an image encoding apparatus 400 drawn according to some embodiments of the present application. In some embodiments, Figure 1 's preprocessing circuit 110 can be implemented by existing circuits in the system, and the preprocessing circuit 110 can be implemented together by the circuits in an image signal processor in the system and the central processing unit 300A in the system. Similar to Figure 3, in this example, the foregoing multiple eigenvalue VF1 to VF3 are calculated by the central processing unit 300A in the system. As Figure 4 shown, the preprocessing circuit 110 includes an automatic exposure controller 412, a buffer 414, a downsampling circuit 416, and an automatic exposure statistics circuit 418, which may be part of the image signal processor, for example, may be an auto exposure module. The automatic exposure controller 412 can control the exposure of the camera 400A according to the relevant parameters stored in the buffer 414. In some embodiments, Figure 1 the operation of the downsampling circuit 112 can be integrated into the downsampling circuit 416 used in the existing automatic exposure module in the system. The operation of the downsampling circuit 416 is similar to Figure 1 that of the downsampling circuit 112, so it will not be elaborated here. The automatic exposure statistics circuit 418 can perform automatic exposure operations according to the luminance values captured by the downsampling circuit 416 to determine the relevant parameters stored in the buffer 414, and transmit information such as the luminance values and the number of pixels to the central processing unit 300A. In this way, the central processing unit 300A can generate the foregoing multiple eigenvalue VF1, VF2, and VF3.

[0034] In Figure 1 , Figure 3 and / or Figure 4 shown in some embodiments, the multiple eigenvalue VF1, VF2, and VF3 are all calculated before entering the weighted prediction circuit 120 and the encoding circuit 130. In this way, the existing circuits of the image signal processor in the system can be used to implement the preprocessing circuit 110, and some operations in the preprocessing circuit 110 can be executed by some circuits in the system. If the multiple eigenvalue VF1, VF2, and VF3 are instead generated by an encoder in the system (for example, the encoding circuit 130), the encoder has to sequentially read out multiple frames F1 to Fn from the memory 100A to perform operations such as downsampling and generating eigenvalue, offset value, etc. (as Figure 2 shown), and read out the frames to be encoded from the memory 100A for encoding. In this way, the required operation timing of the encoder may be too long and does not meet the actual application requirements. In contrast, since the image signal processor is usually the upper circuit (or the previous stage circuit) of the encoder, the operation timing of the image signal processor in processing multiple frames F1 to Fn can be regarded as independent of the operation timing of the encoder in encoding multiple frames F1 to Fn. In this way, downsampling can be performed in advance through the image signal processor to obtain luminance information, so as to provide the encoder with more redundant frame encoding time, thereby being easy to implement in hardware and meeting the timing requirements in actual applications.

[0035] In summary, the image coding device and the weighted prediction image coding method in some embodiments of the present application can use simple operations to determine whether to use weighted prediction operations to code frames. In this way, the required data operation amount and data coupling (or dependency) can be significantly reduced, making it easy to implement in hardware and improving the overall coding speed. On the other hand, in practical applications, the above simple operations can be shared with some circuits in the system. In this way, the related operations of weighted prediction image coding can be integrated into the existing system.

[0036] Although the embodiments of the present application are as described above, these embodiments are not used to limit the present application. Those with ordinary knowledge in the technical field can modify the technical features of the present application based on the explicit or implicit content of the present application. All such modifications may fall within the scope of patent protection sought by the present application. In other words, the scope of patent protection of the present application shall be determined by the scope of the patent application defined in this specification.

[0037]

Symbol Description

[0038] 100, 300, 400: Image coding device

[0039] 100A: Memory

[0040] 110: Preprocessing circuit

[0041] 112: Downsampling circuit

[0042] 114: Feature calculation circuit

[0043] 120: Weighted prediction circuit

[0044] 122: Buffer

[0045] 124: Offset calculation circuit

[0046] 126: Decision circuit

[0047] 130: Coding circuit

[0048] 200: Weighted prediction image coding method

[0049] 300A: Central processing unit

[0050] 312: Sampling circuit

[0051] 314: Accumulation circuit

[0052] 316: Counter

[0053] 400A: Camera

[0054] 412: Automatic exposure controller

[0055] 414: Buffer

[0056] 416: Downsampling circuit

[0057] 418: Automatic exposure statistics circuit

[0058] F1~Fn: Picture frames

[0059] OS1, OS2: Offset values

[0060] S210, S220, S230, S240, S250: Operations

[0061] VC: Control signal

[0062] VF1~VF3: Eigenvalues

[0063] VI: Video

Claims

1. A weighted prediction image coding method, characterized in that, The method includes: Calculating a first eigenvalue of a first picture frame, a second eigenvalue of a second picture frame, and a third eigenvalue of a third picture frame; Calculating a first offset value based on the first eigenvalue and the second eigenvalue, and calculating a second offset value based on the second eigenvalue and the third eigenvalue; And When both the first offset value and the second offset value are greater than or both are less than a predetermined value, encoding the third picture frame using a weighted prediction operation, wherein the first picture frame, the second picture frame, and the third picture frame are a plurality of picture frames in temporal sequence.

2. The weighted prediction image coding method according to claim 1, wherein The calculating the first eigenvalue of the first picture frame, the second eigenvalue of the second picture frame, and the third eigenvalue of the third picture frame includes: Generating a downsampled picture frame based on a corresponding picture frame among the first picture frame, the second picture frame, and the third picture frame; and Determining a corresponding eigenvalue among the first eigenvalue, the second eigenvalue, and the third eigenvalue based on the luminance information of the downsampled picture frame.

3. The weighted prediction image coding method according to claim 2, wherein The determining the corresponding eigenvalue among the first eigenvalue, the second eigenvalue, and the third eigenvalue based on the luminance information of the downsampled picture frame includes: Determining the luminance information of the downsampled picture frame based on the luminance values of at least some pixels in the downsampled picture frame; and Performing a statistical operation based on the luminance information of the downsampled picture frame to generate the corresponding eigenvalue.

4. The weighted prediction image coding method according to claim 1, characterized in that The calculating the first offset value based on the first eigenvalue and the second eigenvalue, and calculating the second offset value based on the second eigenvalue and the third eigenvalue includes: Calculating the difference between the first eigenvalue and the second eigenvalue to determine the first offset value; And Calculating the difference between the second eigenvalue and the third eigenvalue to determine the second offset value.

5. The weighted prediction image coding method according to claim 1, wherein The method includes: When the first offset value and the second offset value are not both greater than or not both less than the predetermined value, not encoding the third picture frame using the weighted prediction operation.

6. An image encoding device, characterized in that, The apparatus includes: A preprocessing circuit that calculates a first eigenvalue of a first picture frame, a second eigenvalue of a second picture frame, and a third eigenvalue of a third picture frame; and A weighted prediction circuit that calculates a first offset value based on the first eigenvalue and the second eigenvalue, and calculates a second offset value based on the second eigenvalue and the third eigenvalue, and controls an encoding circuit to encode the third picture frame using a weighted prediction operation when both the first offset value and the second offset value are greater than or both are less than a predetermined value, where the first picture frame, the second picture frame, and the third picture frame are a plurality of picture frames in temporal sequence.

7. The image encoding apparatus according to claim 6, wherein The preprocessing circuit generates a downsampled picture frame based on a corresponding picture frame among the first picture frame, the second picture frame, and the third picture frame, and determines a corresponding eigenvalue among the first eigenvalue, the second eigenvalue, and the third eigenvalue based on the luminance information of the downsampled picture frame.

8. The image encoding device according to claim 7, characterized in that, The preprocessing circuit includes: A downsampling circuit that downsamples the corresponding picture frame to obtain the downsampled picture frame; and A feature calculation circuit determines the luminance information of the downsampled frame image based on the luminance values of at least some of the pixels in the downsampled frame image, and performs a statistical operation based on the luminance information of the downsampled frame image to determine the corresponding feature value.

9. The image encoding device according to claim 8, wherein The feature calculation circuit calculates an average luminance value of all the pixels in the downsampled frame image to determine the corresponding feature value.

10. The image encoding device according to claim 6, characterized in that, The preprocessing circuit includes: A part of the circuit in an image signal processor that downsamples a corresponding one of the first frame image, the second frame image, and the third frame image to generate a downsampled frame image; and A central processing unit that determines a corresponding feature value among the first feature value, the second feature value, and the third feature value based on the luminance information of the downsampled frame image.

11. The image encoding device according to claim 10, wherein, The part of the circuit is an automatic exposure module or a noise reduction module.

12. The image encoding device according to claim 6, wherein The weighted prediction circuit includes: An offset calculation circuit that calculates a difference between the first feature value and the second feature value to determine the first offset value, and calculates a difference between the second feature value and the third feature value to determine the second offset value; And A decision circuit that controls the encoding circuit to encode the third frame image using the weighted prediction operation when both the first offset value and the second offset value are greater than or both are less than the predetermined value.

13. The image encoding apparatus according to claim 6, wherein When the first offset value and the second offset value are not both greater than or not both less than the predetermined value, the weighted prediction circuit does not control the encoding circuit to encode the third frame image using the weighted prediction operation.

Images