Dynamic image compensation system with correction function and parameter correction method thereof

Abstract

The invention discloses a dynamic image compensation (MEMC) system with a correction function and a parameter correction method thereof, and relates to a parameter correction method for dynamic image compensation, which is used for correcting a dynamic compensation program. The parameter correction method comprises the following steps: storing a default feature set; executing a dynamic compensation program on the corrected image to generate a to-be-detected feature set; generating a correction parameter according to a difference value between the default feature set and the to-be-detected feature set; and adjusting the correction parameter of the dynamic compensation program according to the correction parameter.

Description

Technical Field

[0001] Regarding an electronic system and a processing method for correcting digital images, particularly related to a dynamic image compensation system with a correction function and its parameter correction method. Background Art

[0002] Motion Estimation and Motion Compensation (MEMC) is a technology for image frame interpolation, which can play low image frames at a higher image frame rate, making the image motion smoother.

[0003] With the iterative update of the MEMC algorithm, the updated image processing chips will be applied in new image playback products (such as monitors or playback programs). Since the MEMC algorithms (or the chips installed) used in different generations (or product lines) of image playback products are different, the digital images after compensation for each product will also be different. Summary of the Invention

[0004] In one embodiment, a dynamic image compensation system with a correction function includes a storage unit, a MEMC unit, and a processing unit. The storage unit stores a default feature set of at least one specified tile feature; the MEMC unit receives a corrected image and executes a dynamic compensation program on the corrected image to generate a set of to-be-tested features; the processing unit is connected to the storage unit and the MEMC unit, and generates a correction parameter according to the difference between the default feature set and the set of to-be-tested features; wherein, the MEMC unit is further configured to correct the output result of the dynamic compensation program according to the correction parameter.

[0005] In one embodiment, it further includes an image source, connected to the MEMC unit, and outputs a corrected image to the MEMC unit, wherein the corrected image has at least one frame feature, and the MEMC unit generates a set of to-be-tested features according to at least one frame feature of the corrected image.

[0006] In one embodiment, the image source is an image generator, and the image generator sequentially generates a specified image with changes in specified tile features of different levels of detail according to a tile feature table as the corrected image.

[0007] In one embodiment, the image source is a video and audio player, and the video and audio player outputs a specified image as the corrected image.

[0008] In one embodiment, the MEMC unit obtains at least one specified tile feature from the corrected image, calculates the to-be-tested feature values of each specified tile feature, and records the to-be-tested feature values of each specified tile feature to the set of to-be-tested features.

[0009] In one embodiment, the storage unit stores an information correspondence table, the information correspondence table records a default feature set of at least one specified tile feature, and the processing unit includes a comparison unit and a correction unit. The comparison unit is connected to the MEMC unit and calculates the difference between the measured feature value and the default feature set; the correction unit is connected to the comparison unit and confirms the difference according to a threshold value, wherein when the difference is greater than the threshold value, the correction unit executes a linear regression program according to the measured feature set and the default feature set to generate a correction parameter.

[0010] In some embodiments, the correction unit first executes a linear regression program on the measured feature value method and the default feature set to obtain a temporary feature set, and then executes a least squares equation on the temporary feature set to generate a correction parameter.

[0011] A method for correcting parameters of dynamic image compensation includes storing a default feature set; performing a dynamic compensation program on a corrected image to generate a measured feature set; generating a correction parameter according to the difference between the default feature set and the measured feature set; and adjusting a correction parameter of the dynamic compensation program according to the correction parameter.

[0012] In some embodiments, it further includes connecting to an image source; receiving a corrected image with a specified tile feature provided by the image source, wherein the default detail set includes default feature values of the specified tile feature.

[0013] In some embodiments, it further includes loading a tile feature table; generating a specified image with a specified tile feature having different levels of detail as a corrected image according to the tile feature table.

[0014] In one embodiment, the step of performing a dynamic compensation program on a corrected image according to a compensation parameter to generate a measured feature set includes obtaining at least one specified tile feature from the corrected image; calculating the measured feature values of each specified tile feature; and establishing a measured feature set with the measured feature values of at least one specified tile feature.

[0015] In one embodiment, the default feature set is recorded in the information correspondence table, and the step of adjusting the parameters of the dynamic compensation program according to the correction parameter includes adjusting the information correspondence table according to the correction parameter to generate a corrected correspondence table.

[0016] In one embodiment, it further includes loading a tile feature table; sequentially generating a specified image with at least one specified tile feature having different levels of detail as a corrected image according to the tile feature table.

[0017] In one embodiment, the step of generating a correction parameter according to the difference between the default feature set and the measured feature set includes executing a linear regression program on the measured feature set and the default feature set to obtain a temporary feature set; and executing a least squares equation on the temporary feature set to generate a correction parameter.

[0018] In one embodiment, the step of performing a linear regression program on the set of features to be measured and the default feature set to obtain a temporary feature set includes forming an initial cluster point distribution with the method of feature values to be measured and the default feature set; identifying and removing outliers in the initial cluster point distribution to obtain the temporary feature set.

[0019] In one embodiment, it further includes receiving an input image; performing a dynamic compensation program on the input image according to the calibration parameters to generate an interpolated image.

[0020] In summary, according to any one of the embodiments, the dynamic image compensation system with a calibration function and its parameter calibration method provide an automatic correction process for dynamic compensation technology, thereby generating correction parameters for correcting the output result of the dynamic compensation program during normal operation, so that the MEMC algorithms (or the chips applying them) in different devices can output the same compensation result, thereby reducing the workload of developers, reducing development costs, and also enabling the compensation results of each generation of MEMC algorithms to tend to be consistent after the firmware or software version is updated. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic structural diagram of a dynamic image compensation system with a calibration function according to an embodiment.

[0022] Figure 2 It is a schematic flowchart of a parameter calibration method for dynamic image compensation according to an embodiment.

[0023] Figure 3 For Figure 1 a schematic structural diagram of a demonstration example of the compensation system.

[0024] Figure 4 For Figure 1 a schematic structural diagram of another demonstration example of the compensation system.

[0025] Figure 5A It is a calibrated image of a checkerboard tile with a certain pitch according to an embodiment.

[0026] Figure 5B It is a calibrated image of a checkerboard tile with another pitch according to an embodiment.

[0027] Figure 5C It is a calibrated image of a checkerboard tile with another pitch according to an embodiment.

[0028] Figure 5D It is a calibrated image of a checkerboard tile with another pitch according to an embodiment.

[0029] Figure 5E It is a calibrated image of a checkerboard tile with another pitch according to an embodiment.

[0030] Figure 5F The corrected image of checkerboard tiles with another pitch for an embodiment.

[0031] Figure 5G The corrected image of checkerboard tiles with another pitch for an embodiment.

[0032] Figure 5H The corrected image of checkerboard tiles with another pitch for an embodiment.

[0033] Figure 5I The corrected image of checkerboard tiles with another pitch for an embodiment.

[0034] Figure 6 The schematic diagram of the system architecture for an embodiment.

[0035] Figure 7 The point cloud distribution diagram of the default feature set for the temporary feature set for an embodiment.

[0036] Figure 8 The schematic diagram of the architecture of the compensation system and the display for an embodiment. Detailed implementation manners

[0037] Please refer to Figure 1 , a dynamic image compensation system with a correction function (hereinafter referred to as compensation system 100) is applicable to include a storage unit 110, a motion estimation and motion compensation (Motion Estimation and Motion Compensation, hereinafter referred to as MEMC) unit 120, and a processing unit 130. The processing unit 130 is electrically connected to the storage unit 110 and the MEMC unit 120. In some embodiments, the compensation system 100 can be applied to an image processing chip or a combination thereof with an audio and video player.

[0038] The storage unit 110 stores a default feature set 112. Among them, the default feature set 112 includes multiple preset feature values 113 (golden values). The default feature value 113 is the default feature value 113 of the specified tile feature obtained by passing the target image through the expected MEMC algorithm and is pre-stored in the storage unit 110. Generally speaking, the expected MEMC algorithm means that the compensation effect presented by the image processed by the MEMC algorithm meets the expectations of the designer (for example, the compensation results of the same image by different image processing chips are generally the same).

[0039] In some embodiments, the default feature set 112 can be stored in the storage unit 110 in the form of an information table 111. In other words, the storage unit 110 stores the information table 111 of the target image, and one or more specified tile features of the default target image and the default feature values 113 of the respective specified tile features (i.e., the default feature set 112) are recorded in this information table 111. Further, the storage unit 110 can store multiple target images and their corresponding information tables 111.

[0040] Generally speaking, the target image has at least one frame of image. Among them, the target image can be, for example, a checkerboard image with different spacings, a checkerboard image with different color blocks, or an image with moving objects, etc. Further, the target image can be, but is not limited to, an image with moving objects or a static image. For example, when the target image is a static image, the target image can be a checkerboard image, a landscape painting, an animal painting, or an architectural painting, etc. When the target image is an image with moving objects, the target image can be an image of a sports event, a racing car image, or an animal movement image, etc.

[0041] The MEMC unit 120 has a dynamic compensation program 121. In normal operation (i.e., the general mode), the MEMC unit 120 compensates the input image of the input compensation system 100 by executing the dynamic compensation program 121 to adjust (increase) the number of frames of the output image, and then provides the generated compensated image 830 to the display 820 for display, as shown in Figure 8 shown. Further, the storage unit 110 also stores the dynamic compensation program 121. When the compensation system 100 is started, the MEMC unit 120 reads and loads the dynamic compensation program 121 from the storage unit 110 for subsequent compensation of the input image.

[0042] Here, the dynamic compensation program 121 and the expected MEMC algorithm are different versions of the MEMC algorithm or different types of MEMC algorithms. In some embodiments, the MEMC unit 120 can be a software MEMC algorithm (i.e., the dynamic compensation program 121), or a circuit that runs the MEMC algorithm in hardware.

[0043] To ensure that the application product has a consistent (or expected) compensation effect, the compensation system 100 will first perform calibration of the correction parameters of the dynamic compensation program 121. In other words, the compensation system 100 has an operation mode and a calibration mode.

[0044] Please refer to Figure 1 and Figure 2, in the calibration mode, the MEMC unit 120 receives the calibration image 150, and the MEMC unit 120 performs a dynamic compensation program 121 on the calibration image 150 to generate a set of features to be measured 160 of the calibration image 150 (step S310).

[0045] In some embodiments, the MEMC unit 120 obtains one or more specified pattern features from the calibration image 150, calculates the level of detail of each specified pattern feature to obtain a feature value to be measured, and then forms the set of features to be measured 160 with the feature values to be measured of all the specified pattern features. In other words, the set of features to be measured 160 includes the feature values to be measured of one or more specified pattern features. Among them, the feature value to be measured of each specified pattern feature is a value representing the amount of detail.

[0046] In some embodiments, the specified pattern feature may be the tile detail, tile size, tile edge, color, motion speed or a combination thereof in the calibration image 150. For example, please refer to Figures 5A - 5I , taking the checkerboard image 150A as the calibration image 150 for illustration. At this time, the specified pattern feature may be the spacing of the checkerboard. When performing the dynamic compensation program 121 on the calibration image 150, the MEMC unit 120 can analyze the calibration image 150 to find the checkerboard spacing in the calibration image 150 and calculate the size of this checkerboard spacing to obtain the feature value to be measured. In other implementation aspects, different numbers of tiles can be selected for the calculation process of the dynamic compensation program 121.

[0047] In some embodiments, the MEMC unit 120 is connected to the image source 140. The image source 140 outputs the calibration image 150 to the MEMC unit 120. The calibration image 150 has one or more specified pattern features. Among them, the specified pattern feature is the tile detail, tile size, tile edge, color, motion speed or a combination thereof in the calibration image 150. In other words, the image feature and the specified pattern feature are the same type of feature category. Among them, the image source 140 may be an image generator 141 built into the compensation system 100 (as Figure 3 shown) or an image player 142 external to the compensation system 100 (as Figure 4 shown). That is to say, the calibration image 150 received by the MEMC unit 120 can be generated by the image generator 141 built into the compensation system 100 itself (as Figure 3 shown), or can be from outside the compensation system 100 (as Figure 4 shown).

[0048] In some embodiments, refer to Figure 3, the compensation system 100 may further include an image generator 141, and this image generator 141 is connected to the MEMC unit 120. The image generator 141 is used to generate a corrected image 150 with specified tile features. Specifically, the storage unit 110 also stores a tile feature table (not drawn). The image generator 141 will load the tile feature table and generate a specified image with specified tile features at different levels of detail according to the loaded tile feature table as the corrected image 150.

[0049] For example, the image generator 141 generates a frame-specified picture with specified tile features, then determines the selected specified tile features according to the tile feature table, and then sequentially adjusts the detail levels of the selected specified tile features in this frame-specified picture to generate multiple frame-specified pictures with different detail levels. Finally, the image generator 141 uses the specified image composed of these multiple frame-specified pictures as the corrected image 150 and outputs it to the MEMC unit 120. For example: The image generator 141 can generate multiple sets of corrected images 150, and each set of corrected images 150 is the default pattern of a checkerboard image 150A with different spacings, color blocks or combinations thereof, such as Figures 5A - 5I shown. Figures 5A - 5I are multiple checkerboard images 150A with different spacings (i.e., the corrected image 150). In some embodiments, this tile feature table may record the correspondence between one or more specified tile features mapped to each specified image and their related generation parameters.

[0050] In some embodiments, referring to Figure 4 , the MEMC unit 120 may be connected to an image player 142 outside the compensation system 100. Here, the image player 142 can output multiple frame-specified pictures (i.e., the corrected image 150) with specified tile features at different levels of detail to the MEMC unit 120. For example, taking the picture movement speed as the specified tile feature, after the MEMC unit 120 receives the specified picture output by the image player 142, the MEMC unit 120 will calculate the picture movement speed of each frame-specified picture as the measured feature value. In some embodiments, the image player 142 can be a personal computer, an optical drive, or other multimedia players. The image player 142 can provide different types of corrected images 150. For example, the image player 142 can play images of sports events, racing images, or animal movement images, etc., as the corrected image 150.

[0051] Next, the MEMC unit 120 transmits the set of features to be measured 160 to the processing unit 130. In addition, the processing unit 130 loads the default feature set 112 that has been pre-stored in the storage unit 110, and accordingly obtains the corresponding default feature set 112 (step S320). Then, the processing unit 130 calculates the difference between the default feature set 112 and the set of features to be measured 160, and the processing unit 130 generates correction parameters based on each difference.

[0052] Specifically, the processing unit 130 calculates the difference between each feature value to be measured in the set of features to be measured 160 and the corresponding default feature value 113 in the default feature set 112, and determines whether correction is required based on the calculated difference. If any difference is too large (e.g., greater than or equal to the threshold), the processing unit 130 further generates correction parameters based on the default feature set 112 and the set of features to be measured 160 (step S330). Conversely, if the differences all meet the standard (e.g., less than the threshold), it means that the compensation result output by the MEMC unit 120 conforms to the expectation, and correction does not need to be performed (i.e., the processing unit 130 does not generate correction parameters, and the compensation system 100 completes the correction. Finally, the processing unit 130 adjusts the correction parameters of the dynamic compensation program 121 according to the generated correction parameters.

[0053] For example, taking Figures 5A - 5I the checkerboard image 150A (i.e., the calibration image 150) as an example, referring to Figure 1 and Figures 5A - 5I , the image source 140 selects "detail" as the variable of the calibration image 150 (i.e., specifies the tile feature). For the checkerboard image 150A, the degree of detail can be the size of the checkerboard spacing. The image source 140 outputs checkerboard images 150A with different checkerboard spacings to the MEMC unit 120. The MEMC unit 120 calculates the "amount" of "detail" in the calibration image 150 according to the MEMC algorithm carried, and outputs the calculated detail amount (i.e., the feature value to be measured) to the processing unit 130. The processing unit 130 then compares the calculated detail amount with the corresponding target amount (i.e., the default feature value 113) to determine whether correction is required.

[0054] It should be clearly understood that in some embodiments, after step S330, the compensation system 100 can confirm whether the correction is effective by repeatedly executing steps S310 to S330. In other words, the compensation system 100 repeatedly executes steps S310 to S330 until the differences of the calibration image 150 all meet the standard.

[0055] In some embodiments, the calibration image 150 and the target image that the processing unit 130 compares with each other (i.e., calculates the difference between the respective default feature sets 112 and the to-be-measured feature sets 160) may be images of the same type. For example, both the calibration image 150 and the target image are checkerboard images.

[0056] In some embodiments, the calibration parameters may be recorded in the correction correspondence table 122. Specifically, after the processing unit 130 generates the calibration parameters for each specified tile feature, the generated calibration parameters are recorded in the correction correspondence table 122 as correction parameters, and then the correction correspondence table 122 is output to the MEMC unit 120. The MEMC unit 120 can then correct the output result of the dynamic compensation program 121 according to the correction correspondence table 122. In some embodiments, the correction correspondence table 122 may be stored in the storage unit 110. When the MEMC unit 120 executes the dynamic compensation program 121, the MEMC unit 120 reads the correction correspondence table 122 from the storage unit 110 and loads it into the temporary memory of the MEMC unit 120 for correction.

[0057] In some embodiments, the correction correspondence table 122 and the information correspondence table 111 may be independent of each other. In other embodiments, the correction correspondence table 122 and the information correspondence table 111 may also be integrated into a single form. In a demonstration example, the information correspondence table 111 has default feature values 113 and correction parameters for one or more specified tile features. After the processing unit 130 generates the calibration parameters, the generated calibration parameters are used to adjust the corresponding correction parameters in the information correspondence table 111 to form a corrected information correspondence table, and then the corrected information correspondence table replaces (or updates) the information correspondence table 111 stored in the storage unit 110. In another demonstration example, after the processing unit 130 generates the calibration parameters, the generated calibration parameters are recorded in the information correspondence table 111 as correction parameters.

[0058] In some embodiments, please refer to Figure 6 , the processing unit 130 includes a comparison unit 131 and a calibration unit 132. The comparison unit 131 is connected to the MEMC unit 120 and the storage unit 110. The calibration unit 132 is connected to the comparison unit 131 and is externally connected to the MEMC unit 120 and / or the storage unit 110.

[0059] The comparison unit 131 receives the eigenvalues to be measured calculated by the MEMC unit 120, and calculates the differences between each eigenvalue to be measured and the corresponding default eigenvalue 113 in the default feature set 112. Then, the comparison unit 131 outputs the differences corresponding to the eigenvalues to be measured to the calibration unit 132. The calibration unit 132 determines whether the differences corresponding to each eigenvalue to be measured are greater than the threshold. If the difference is greater than or equal to the threshold, the calibration unit 132 executes a linear regression program based on the default feature set 112 and the default feature set 112, and generates calibration parameters for each eigenvalue to be measured according to the execution result. If the difference is less than the threshold, the calibration unit 132 does not generate calibration parameters.

[0060] In some embodiments, the calibration unit 132 may select some eigenvalues from the eigenvalue set 160 to be measured for generation. In particular, the outliers (unlabeled) in the eigenvalue set 160 to be measured are excluded. Thereby, the influence of the outliers on the compensation result is reduced. Specifically, the calibration unit 132 executes a linear regression program on the eigenvalue set 160 to be measured and the default feature set 112 to obtain the initial cluster point distribution of the default eigenvalue 113 with respect to the eigenvalue to be measured. Among them, the initial cluster point distribution is composed of multiple cluster points, and each cluster point is composed of an eigenvalue to be measured with a specified tile feature and its corresponding preset eigenvalue 113. After the calibration unit 132 removes the outliers in the initial cluster point distribution, a temporary cluster point distribution composed of the remaining cluster points is formed (as Figure 7 shown), and a temporary feature set is obtained therefrom. In the temporary cluster point distribution, the eigenvalues to be measured corresponding to these remaining cluster points are the temporary feature set. Then, the calibration unit 132 calculates the regression line fitted to the temporary cluster point distribution (such as the thick black line shown in Figure 7 ) and obtains the calibration parameters accordingly. In some embodiments, the outliers can be determined by standard deviation, quartile, or box plot, etc.

[0061] In some embodiments, the calibration unit 132 fits a set of regression lines (such as the thick black line shown in Figure 7 ) that conform to the temporary cluster point distribution and its corresponding linear equation through linear regression with the least squares method. Specifically, the calibration unit 132 uses the least squares method to calculate the linear equation of the regression line fitted to the temporary cluster point distribution, and takes the coefficients of the linear equation as the calibration parameters.

[0062] Taking the Figure 7 cluster point distribution as an example, Figure 7 the X-axis in represents the preset eigenvalue 113, and the Y-axis represents the eigenvalue to be measured. And the dots in Figure 7 represent the differences mentioned above, and each difference can correspond to a preset eigenvalue 113 on the X-axis and an eigenvalue to be measured on the Y-axis.

[0063] The algorithm of the least squares equation is shown in Equation 1. Equation 1 can derive the calculation formulas for the calibration parameters (α, β) through partial differentiation, as shown in Equations 2 and 3. Substitute the numerical values of each group of points in Figure 7 into Equations 2 and 3, that is, fill in the preset eigenvalue 113 for y_i and fill in the to-be-measured eigenvalue for x^_i. Therefore, α can be calculated as 1.63 and β as 130.17. That is to say, for the distribution of the group of points shown in Figure 7 , the calibration unit 132 fits the regression line represented by Equation 4 through linear regression combined with the least squares method. Therefore, the calibration unit 132 can obtain the correction parameters as (α, β) = (1.63, 130.17).

[0064]

[0065]

[0066]

[0067] y(i)=1.63x i +130.17 Equation 4

[0068] y = αx + β Equation 5

[0069] In some embodiments of step S330, when obtaining the correction parameters (α, β), the calibration unit 132 will first temporarily store the obtained correction parameters (α, β) in the storage unit 110, and then repeatedly calculate the differences corresponding to other specified tile features and calculate the corresponding correction parameters (α, β) accordingly until all the correction parameters (α, β) corresponding to all specified tile features are obtained. Then, the calibration unit 132 aggregates all the correction parameters (α, β) into a correction correspondence table 122, and then stores the correction correspondence table 122 in the storage unit 110.

[0070] After the compensation system 100 receives a new input image. In normal operation (i.e., the general mode), the MEMC unit 120 will load the correction correspondence table 122. The MEMC unit 120 corrects the output result of the dynamic compensation program 121 according to the correction correspondence table 122. In other words, during the execution of the dynamic compensation program 121 for the input image, the MEMC unit 120 will first calculate the estimated feature values of each specified tile feature of each frame in this input image, and then correct the estimated feature values of the corresponding specified tile features with the correction parameters in the correction correspondence table 122 to obtain the compensated feature values (i.e., the set of compensation parameters) of each specified tile feature, and then generate a compensated image 830 containing at least one interpolated frame according to the input image and the set of compensation parameters.

[0071] In some embodiments, the MEMC unit 120 corrects the output result of the dynamic compensation program 121 with the correction algorithm shown in Equation 5. In Equation 5, x represents the initially estimated eigenvalue, α and β are the corresponding correction parameters in the correction look-up table 122, and y represents the compensated eigenvalue. For example, during the process of the MEMC unit 120 executing the dynamic compensation program 121 on the input image, the MEMC unit 120 substitutes the obtained initially estimated eigenvalue into x in Equation 5 and substitutes the corresponding correction parameters (α, β) in the correction look-up table 122, and then can calculate the compensated eigenvalue (y). Then, the MEMC unit 120 generates an interpolated image according to the compensated eigenvalue (y) and the input image and inserts it into the input image to form the compensated image 830. As Figure 8 shown, the audio-visual player 810 (i.e., the image source 140) provides the corrected image 150 to the compensation system 100. The compensation system 100 corrects the corrected image and generates the compensated image 830. The compensation system 100 outputs the compensated image 830 to the display 820.

[0072] In summary, according to any one of the embodiments, the compensation system 100 or its parameter correction method provides an automatic correction function for the dynamic compensation technology, so as to generate correction parameters for correcting the output result of the dynamic compensation program 121 during normal operation, so that the MEMC algorithms (or their chips) in different devices can output the same compensation result, thereby reducing the workload of developers, reducing the development cost, and also enabling the compensation results of each generation of MEMC algorithms to be consistent after the firmware or software version is updated.

[0073]

Symbol Explanation

[0074] 100: Compensation system

[0075] 110: Storage unit

[0076] 111: Information look-up table

[0077] 112: Default feature set

[0078] 113: Preset eigenvalue

[0079] 120: MEMC unit

[0080] 121: Dynamic compensation program

[0081] 122: Correction look-up table

[0082] 130: Processing unit

[0083] 131: Comparison unit

[0084] 132: Calibration unit

[0085] 140: Image source

[0086] 141: Image generator

[0087] 142: Image player

[0088] 150: Calibrated image

[0089] 150A: Checkerboard image

[0090] 160: Feature set to be measured

[0091] 810: Audio and video player

[0092] 820: Display

[0093] 830: Compensated image

[0094] S310, S320, S330, S340: Steps

Claims

1. A dynamic image compensation system with correction function, comprising: A storage unit for storing a default feature set of at least one specified tile feature; A motion estimation and motion compensation (MEMC) unit for receiving a corrected image and performing a dynamic compensation program on the corrected image to generate a to-be-tested feature set; And A processing unit connected to the storage unit and the MEMC unit for generating a correction parameter according to the difference between the default feature set and the to-be-tested feature set; Wherein, the MEMC unit is further configured to correct the output result of the dynamic compensation program according to the correction parameter.

2. The dynamic image compensation system with correction function according to claim 1, further comprising an image source connected to the MEMC unit for outputting the corrected image to the MEMC unit, wherein the corrected image has at least one frame feature, and the MEMC unit generates the to-be-tested feature set according to the at least one frame feature of the corrected image.

3. The dynamic image compensation system with correction function according to claim 2, wherein the image source is an image generator, and the image generator sequentially generates a specified image with changes in the specified tile features of different levels of detail according to a tile feature table as the corrected image.

4. The dynamic image compensation system with correction function according to claim 1, wherein the MEMC unit obtains at least one of the specified tile features from the corrected image, calculates the to-be-tested feature values of each of the specified tile features, and records the to-be-tested feature values of each of the specified tile features to the to-be-tested feature set; The storage unit stores an information correspondence table, and the information correspondence table records the default feature set of the at least one specified tile feature, and The processing unit includes: A comparison unit connected to the MEMC unit for calculating the difference between the to-be-tested feature value and the default feature set; and A correction unit connected to the comparison unit for confirming the difference according to a threshold value. When the difference is greater than the threshold value, the correction unit performs a linear regression program according to the to-be-tested feature set and the default feature set to generate the correction parameter. Wherein, the correction unit first performs the linear regression program on the to-be-tested feature value method and the default feature set to obtain a temporary feature set, and then performs a least squares equation on the temporary feature set to generate the correction parameter.

5. A parameter correction method for dynamic image compensation, comprising: Storing a default feature set; Performing a dynamic compensation program on a corrected image to generate a to-be-tested feature set; Generating a correction parameter according to the difference between the default feature set and the to-be-tested feature set; And Adjusting the correction parameter of the dynamic compensation program according to the correction parameter.

6. The parameter correction method for dynamic image compensation according to claim 5, further comprising: Loading a tile feature table; And Generating a specified image with at least one specified tile feature of different levels of detail according to the tile feature table as the corrected image.

7. The method for parameter correction of dynamic image compensation according to claim 5, wherein the step of performing the dynamic compensation program on the corrected image to generate the set of features to be measured includes: Obtaining at least one specified tile feature from the corrected image; Calculating the feature values to be measured of each of the specified tile features; And Establishing the set of features to be measured with the feature values to be measured of the at least one specified tile feature.

8. The method for parameter correction of dynamic image compensation according to claim 5, wherein the step of generating the correction parameter according to the difference between the default feature set and the set of features to be measured includes: Performing a linear regression program on the set of features to be measured and the default feature set to obtain a temporary feature set; And Performing a least squares program on the temporary feature set to generate the correction parameter.

9. The method for parameter correction of dynamic image compensation according to claim 8, wherein the step of performing the linear regression program on the set of features to be measured and the default feature set to obtain the temporary feature set includes: Forming an initial cluster point distribution with the set of features to be measured and the default feature set; And Identifying and removing the outliers in the initial cluster point distribution to obtain the temporary feature set.

10. The method for parameter correction of dynamic image compensation according to claim 5, further including: Receiving an input image; And Performing the dynamic compensation program on the input image according to the correction parameter to generate a compensated image.

Images