Method for detecting fixed pattern in moving image and system for executing fixed pattern detection

Abstract

The invention relates to a method for detecting a fixed pattern in a dynamic image and a system for executing fixed pattern detection. The invention discloses a method and system for detecting fixed patterns in a dynamic image, and the method comprises the steps: firstly obtaining a continuous dynamic image, and then obtaining a frame image with multi-level resolution from the dynamic image, so as to detect a plurality of fixed patterns in the frame image with the multi-level resolution according to the features of a front frame image and a rear frame image; after a first fixed pattern with a higher resolution is converted into a fixed pattern with a lower resolution, a second fixed pattern is merged with the same lower resolution, and a merged fixed pattern is obtained. And then the combined fixed pattern under the lower resolution is subjected to erosion treatment and then converted into a pattern with the higher resolution, boundary detection is carried out, the first fixed pattern with the higher resolution can be subjected to erosion treatment according to a boundary detection result, and a clear fixed pattern is obtained.

Description

Technical Field

[0001] The specification discloses a method for detecting fixed patterns in dynamic images, and in particular a method and system for detecting fixed patterns in dynamic images by obtaining clear fixed patterns based on frame image features at multi-level resolution. Background Technology

[0002] In the development of displays, one of the most important goals is to improve display quality to enhance the user's video viewing experience. Improvements include enhancing video smoothness and clarity. Existing technology uses a motion estimation and motion compensation (MEMC) technique to compensate for dynamic images in videos, thereby making the performance of dynamic images smoother when displaying videos.

[0003] For reference Figure 1 This diagram illustrates the operational concept of Motion Estimation and Motion Compensation (MEMC) technology. The diagram shows two consecutive frames in a video, represented as frame N-1 (101) and frame N (102), where a moving object 100 is positioned differently in the two frames. The operating principle of MEMC technology is to obtain the correlation of object movement in consecutive images. As shown, the processing circuit calculates the motion vector of object 100 from frame N-1 (101) to frame N (102) based on the object's position in the preceding and following frames. Then, one or more compensation images 105, including the interpolated object 100', are inserted between frame N-1 (101) and frame N (102) using image interpolation technology based on the motion vector of object 100. Thus, by using MEMC technology, the frame rate of the video can be increased, and the moving object 100 also generates the interpolated object 100', making the video appear smoother.

[0004] However, for moving objects, the motion estimation and motion compensation techniques are limited by the range of motion that can be processed based on computing power. If the range of motion exceeds the computing power, errors may occur. Conversely, for fixed patterns in an image, a protection mechanism is required. Otherwise, the motion estimation and motion compensation techniques may cause the originally static fixed pattern to become unstable due to misjudgment, such as the fixed pattern adhering to the background image or the fixed pattern having a halo at the boundary.

[0005] For example, when an image player in a display plays a streaming video downloaded from a streaming platform, there is a hardware processor and detection software in the image player that detects logos in the video. One of the purposes of detecting logos is to separate the stationary logos from the video, so as to avoid the original stationary logos being affected by motion estimation and motion compensation.

[0006] However, existing marker detection techniques may have problems such as the marker pattern being too wide and exceeding the original marker pattern range due to the low resolution of the marker pattern itself; or when using edge-finding methods to obtain the boundary of the marker pattern in high-resolution images, the part inside the boundary of the marker pattern may be misjudged as a non-marker pattern, resulting in marker pattern drift during motion compensation. Summary of the Invention

[0007] To address the problem that existing motion estimation and motion compensation (MEMC) methods cannot adequately handle fixed patterns in dynamic images, this disclosure proposes a method and system for detecting fixed patterns in dynamic images.

[0008] In the fixed pattern detection method in dynamic images, a continuous dynamic image is first acquired, and multi-resolution frame images are obtained from the dynamic image. Multiple fixed patterns in the multi-resolution frame images can be detected based on the features of the preceding and following frames, including at least a first fixed pattern with higher resolution and a second fixed pattern with lower resolution. Next, the first fixed pattern with higher resolution is converted into a converted fixed pattern with lower resolution. This converted fixed pattern and the second fixed pattern are then merged at the same lower resolution to obtain a merged fixed pattern. The merged fixed pattern at lower resolution is then subjected to erosion processing. After the eroded merged fixed pattern is converted into a higher resolution pattern, boundary detection is performed. Based on the boundary detection results, the first fixed pattern with higher resolution is then subjected to erosion processing to obtain a clear fixed pattern.

[0009] Furthermore, in the step of detecting multiple fixed patterns in frame images with multi-level resolution, a matrix scanning window can be used to scan the pixel values ​​of each frame image, and the fixed patterns at each resolution can be detected based on the absolute difference between the pixel values ​​of consecutive frames.

[0010] Furthermore, in the step of detecting multiple fixed patterns in frame images with multi-level resolution, the highest and lowest resolution frame images are discarded, thereby saving computing power.

[0011] Furthermore, the merged fixed pattern after erosion can be converted into a higher resolution fixed pattern by repeating pixels, and the boundary of the merged fixed pattern at high resolution can be detected based on the pixel value gradient changes.

[0012] For erosion processing of merged fixed patterns with lower resolution, image processing is used to remove the boundary pixels of the merged fixed patterns. In frame images with multiple resolutions obtained from self-moving images, when erosion processing is performed on the merged fixed patterns at lower resolution, the erosion processing is further performed based on the boundary detection results of a lower-resolution merged fixed pattern obtained from a lower resolution level and then converted to a lower resolution.

[0013] In the implementation plan, after obtaining a clear fixed pattern, motion estimation and motion compensation can be further performed on the dynamic image based on the clear fixed pattern, with the aim of improving the frame rate of the dynamic image.

[0014] Furthermore, when a clear fixed pattern is obtained, the pixels of each frame that are judged to be clear fixed patterns are set as flag "1" and recorded in the memory. When motion estimation and motion compensation are performed on the dynamic image, the pixels that have been set as flag "1" can be ignored based on the memory record.

[0015] Furthermore, in the process of setting the markers for each frame pixel of a clear fixed pattern, the historical records stored in the memory are further referenced to assess the confidence level of the clear fixed pattern.

[0016] To further understand the features and technical content of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are for reference and illustration only and are not intended to limit the present invention. Attached Figure Description

[0017] Figure 1 This diagram illustrates the operational concept of existing motion estimation and motion compensation technologies.

[0018] Figure 2 An example diagram showing a logo pattern in a dynamic image;

[0019] Figure 3 This diagram illustrates the halo effect produced by motion estimation and motion compensation in a moving image of a logo pattern.

[0020] Figure 4 This diagram illustrates how the hollow effect is created in a logo pattern in a dynamic image through motion estimation and motion compensation.

[0021] Figure 5 A diagram illustrating an embodiment of a system for detecting fixed patterns;

[0022] Figure 6A schematic diagram illustrating an embodiment of a fixed pattern logo in an image;

[0023] Figure 7 This diagram shows an embodiment of the motion estimation and motion compensation system architecture.

[0024] Figure 8 An example diagram showing a fixed pattern merging process performed between different resolution levels;

[0025] Figure 9 An example diagram showing fixed pattern boundary processing between different resolution levels is displayed;

[0026] Figure 10 A flowchart illustrating an embodiment of a method for detecting fixed patterns in dynamic images;

[0027] Figure 11 An embodiment diagram showing a fixed pattern processing method between multiple resolution levels; and

[0028] Figure 12 This displays an example diagram for calculating the sum of absolute differences. Detailed Implementation

[0029] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can understand the advantages and effects of the present invention from the content disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the concept of the present invention. Furthermore, the accompanying drawings of the present invention are for simple illustrative purposes only and are not depictions of actual dimensions; this is stated beforehand. The following embodiments will further describe the relevant technical content of the present invention in detail, but the disclosed content is not intended to limit the scope of protection of the present invention.

[0030] It should be understood that while terms such as "first," "second," and "third" may be used in this document to describe various components or signals, these components or signals should not be limited by these terms. These terms are primarily used to distinguish one component from another, or one signal from another. Furthermore, the term "or" as used herein should, as appropriate, include any combination of one or more of the related listed items.

[0031] This disclosure proposes a method and system for detecting fixed patterns in dynamic images. The method operates within a system that processes image data. The system can be a circuit system in a display device that displays digital streaming images. One of the technical objectives is to address the problem of improper handling of fixed patterns during the process of improving the image frame rate through motion estimation and motion compensation (MEMC) when the circuit system processes dynamic images. For example, when the dynamic image is a low-resolution image (such as a 240*135 pixel image), the fixed pattern (such as a logo) may form a halo, causing it to adhere to the background image. Alternatively, when performing motion estimation and motion compensation on a high-resolution low-resolution image (such as a 480*270 pixel or higher image), although the halo effect of the logo is reduced, it may still cause holes to appear in the part within the boundary of the logo.

[0032] Figure 2 An embodiment diagram showing a logo pattern in a dynamic image is shown. The first frame 21 and the second frame 22 of the dynamic image show a moving object 200 and a fixed logo pattern 210. After motion estimation and motion compensation processing, an interpolated frame 20 is generated based on the motion vector of the object 200. The interpolated object 200' is formed after compensation, and it also includes the logo pattern 210' (LOGO) in a fixed position.

[0033] After motion estimation and motion compensation processing, the fixed pattern (such as...) in the dynamic image... Figure 2 The logo displayed in the upper right corner of each frame (200: "LOGO") may exhibit a halo effect due to low resolution. Figure 3 The illustrated diagram shows a halo image formed around the logo pattern 30 ("LOGO"). If a high-resolution image is processed using motion estimation and motion compensation techniques, and edge-finding techniques are used to find the boundaries of the logo image, it will result in an image like... Figure 4 The logo pattern 30' shown is judged to be a non-logo pattern within the boundary of the "LOGO", resulting in a hollow hole image. The part within the boundary is filled by the background image after motion compensation.

[0034] To address the aforementioned issues, the proposed method for detecting fixed patterns in dynamic images employs multi-level resolution images to detect fixed patterns within the dynamic image. Only after confirming the coverage area of ​​the fixed pattern is motion estimation and motion compensation performed. This method can effectively improve the problems caused by motion estimation and motion compensation in low-resolution or high-resolution images.

[0035] First refer to Figure 5The diagram shows an embodiment of a system for detecting fixed patterns in dynamic images. The system can be a circuit system for processing image data, such as a circuit or module in an image processing device in a television set, which includes processing circuitry and executes the method for detecting fixed patterns in dynamic images in hardware.

[0036] The figure shows that after receiving a dynamic image, the system acquires the previous frame data 501 and the current frame data 502, and uses a fixed pattern detection unit 505, implemented in software or hardware, to determine the fixed pattern in the dynamic image based on the characteristics of the previous and next frames. It is also worth mentioning that, depending on the system's hardware processing capabilities, multiple fixed patterns in frames with different resolutions can be detected simultaneously or in a time-division manner.

[0037] For example, the fixed pattern detection unit 505 can use a Sobel operator to detect the boundary features in the previous frame data 501 and the current frame data 502. It can use a scanning window of a specific size to obtain boundary information (such as pixel gradient changes) in various directions in the image. After determining the boundary, it can then compare the two frames to determine the fixed pattern.

[0038] Subsequently, the fixed pattern detection unit 505 can mark the pixels determined to be fixed patterns to form a fixed pattern mark 507. During this process, the historical record 509 can also be referenced to determine the position of the fixed pattern. According to an embodiment, the historical record 509 can use a memory to record previously determined fixed patterns. By means of the historical record 509, the position where the fixed pattern (such as a mark pattern) is frequently displayed in the dynamic image can be determined, thereby assessing the confidence level of the currently determined fixed pattern.

[0039] The embodiment diagram of the fixed pattern mark 507 can be referred to. Figure 6 The diagram shows the area where a fixed pattern 600 is located in frame 60, determined using fixed pattern detection technology. This fixed pattern 600 is identified as a fixed pattern in the moving image. Pixels identified as fixed patterns can be marked with a flag "1" (Flag=1). Thus, when motion estimation and motion compensation are performed on frame 50, pixels marked with flag "1" can be ignored. As described above, in the process of setting the flags for pixels of each frame with clear fixed patterns, historical records stored in memory can be further referenced to assess the confidence level of clear fixed patterns.

[0040] Furthermore, the fixed pattern detection unit 505 for detecting fixed patterns can be configured in software or hardware within the system architecture that performs motion estimation and motion compensation, as can be referred to. Figure 7 This diagram shows an embodiment of the motion estimation and motion compensation system architecture.

[0041] The figure shows the fixed pattern detection unit 505 described in the above embodiment. The fixed pattern detection unit 505 can be hardware in a circuit system or software that performs fixed pattern detection. It receives dynamic image data 701 from the image source. In addition to performing fixed pattern detection, the dynamic image data 701 is first stored in the buffer 702.

[0042] The fixed pattern detection unit 505 primarily identifies fixed patterns in a dynamic image by comparing consecutive frames, such as a fixed, unchanging logo displayed while watching a video. One method involves designing a matrix scanning window (e.g., a 3x3 pixel matrix). This window scans the pixel values ​​of each frame, and the fixed pattern is detected at different resolutions based on the sum of absolute differences (SAD) between the pixel values ​​of consecutive frames. It's worth noting that because the system can store historical records of fixed patterns, it can also determine the frequently occurring locations of these patterns based on the historical records, allowing the matrix scanning window to be used only to scan the locations where clear, fixed patterns frequently appear.

[0043] For example, you can refer to Figure 12 This diagram illustrates the calculation of absolute differences based on frame pixel values. It shows the pixel values ​​of the first window 11 obtained by scanning the previous frame with a matrix scanning window, and the pixel values ​​of the second window 12 obtained by scanning the current frame. The absolute values ​​of the corresponding positions in the previous and current frames within the window are subtracted and the result is the absolute difference displayed in the third window 13.

[0044] Next, the sum of the absolute differences is displayed in the third window 13. The legend shows that the sum of absolute differences (SAD) is 95 (1+2+1+0+4+34+38+12+3=95). This can be compared with the threshold designed by the system. For example, the number of pixels belonging to a fixed pattern can be counted. Those exceeding a certain threshold are considered fixed patterns. This is used to evaluate whether the window covers the features of a fixed pattern. If the threshold is not reached, the fixed pattern can be removed without further processing, thus saving system computing power.

[0045] Furthermore, after the fixed pattern is obtained by the fixed pattern detection unit 505, the pixel values ​​of the fixed pattern are statistically analyzed to obtain statistical information 706, which is provided to the system processing unit 708 and can be used as a historical record for determining the fixed pattern in the future.

[0046] After the fixed pattern is detected in the dynamic image data 701 by the fixed pattern detection unit 505, the procedure for obtaining a clear fixed pattern 703 is executed. The clear fixed pattern can be obtained by using the information obtained by the motion compensation estimation unit 707 when performing motion estimation on the continuous frames of the dynamic image data 701, and the information obtained by the processing unit 708 based on historical records, through the processing method of multi-level resolution fixed pattern. This can eliminate the halo phenomenon after motion compensation processing caused by the fixed pattern resolution being too low, or the problem of misjudging dynamic objects as fixed patterns due to low (insufficient) resolution, or the problem of holes within the boundary of the fixed pattern caused by excessively high resolution.

[0047] After obtaining a clear fixed pattern, it can be output to the motion compensation frame rate conversion unit 704 for motion compensation. According to the embodiment, one or more compensation images are inserted between two consecutive frames in the dynamic image based on the moving objects therein, thereby increasing the frame rate to obtain a smoother dynamic image. Furthermore, during the processing, parts that have been identified as fixed patterns can be excluded, thereby retaining the clear fixed pattern 703 obtained by the fixed pattern detection unit 505, and finally outputting the frame rate converted image data 705.

[0048] Given the deficiencies in current technology, fixed patterns in dynamic images may have different sizes and resolutions. Therefore, under the processing of relevant circuit systems, the fixed pattern detection method proposed in this disclosure employs a multi-level resolution processing method to obtain clear fixed patterns. (See reference...) Figure 8 The diagram shows an embodiment of fixed pattern merging between different resolution levels, and... Figure 9 An example diagram showing the fixed pattern boundary processing between different resolution levels is displayed.

[0049] according to Figure 8 In the illustrated embodiment, firstly, frame images with multiple resolutions are acquired from the dynamic image. Then, the fixed pattern detection unit described above performs a second resolution fixed pattern detection 801 to acquire a fixed pattern with a lower resolution (e.g., 1K). On the other hand, the fixed pattern with a higher resolution (e.g., 2K) is converted to a fixed pattern with a second resolution through a step of reducing the first resolution to the second resolution 805. Finally, a merging calculation 803 is performed on two images with the same lower resolution (second resolution) to obtain a merged fixed pattern 807.

[0050] exist Figure 9 The diagram shows that when a first-resolution fixed pattern is obtained by performing a first-resolution merging 901 at a higher resolution, the first fixed pattern with a higher resolution can be eroded based on the boundary detection results of the lower-resolution fixed pattern to obtain a clear fixed pattern.

[0051] This example illustrates a second-resolution erosion 905 performed on a lower-resolution merged fixed pattern, as shown in the figure. This primarily involves image processing to remove boundary pixels of the merged fixed pattern. Next, in the step of upgrading from the second resolution to the first resolution 907, the eroded merged fixed pattern is converted to a higher-resolution fixed pattern by repeating pixels. Then, the first-resolution boundary detection 909 step is performed, where the boundaries of the merged fixed pattern at high resolution are detected based on pixel value gradient changes. Following this, after edge reduction operations at the lower resolution (second resolution), the system can re-add the higher-resolution (first resolution) boundaries. Finally, the higher-resolution first-resolution erosion 903 is performed to obtain a clear fixed pattern.

[0052] The overall process can be referenced. Figure 10 The flowchart illustrates an embodiment of the fixed pattern detection method in the shown dynamic image.

[0053] At the beginning of the process, continuous dynamic images are acquired (step S101), and multi-resolution frame images are obtained through resolution conversion (step S103) to detect multiple fixed patterns corresponding to different resolutions in the multi-resolution dynamic images (step S105). Multiple fixed patterns in the multi-resolution frame images are detected based on the features of preceding and following frame images. One method is to use a Sobel operator to find edges by analyzing the gradient changes of pixel values ​​in each frame image, obtaining the results of the fixed pattern's boundaries in various directions (vertical, horizontal, and diagonal), and then comparing the image features of preceding and following frames to determine the boundaries of the fixed patterns. The scheme for obtaining fixed patterns through multi-resolution includes at least a first fixed pattern with higher resolution and a second fixed pattern with lower resolution.

[0054] According to the description of the above embodiments, a matrix scanning window can be used to scan the pixel values ​​of each frame of the image, and a fixed pattern at each resolution can be detected based on the absolute difference between the pixel values ​​of consecutive frames. The technical concept is that, for each scanned area, in the technique of obtaining a fixed pattern at multiple resolutions, if no fixed pattern is determined from the high-resolution image, even if a fixed pattern is determined from the low-resolution image, it is not considered a fixed pattern overall; conversely, if a fixed pattern is determined from the high-resolution image, but no fixed pattern is determined from the low-resolution image, it is still not considered a fixed pattern overall.

[0055] Once the fixed patterns at each resolution are obtained, the first fixed pattern with a higher resolution can be converted into a converted fixed pattern with a lower resolution (step S107). Then, the converted fixed pattern with the same lower resolution can be merged with the second fixed pattern obtained above to obtain a merged fixed pattern (step S109). In this way, the merged fixed pattern with the lower resolution can be subjected to erosion processing (step S111).

[0056] Next, the merged fixed pattern after erosion processing is converted to a higher resolution. One method is to convert the merged fixed pattern after erosion processing to a higher resolution fixed pattern by repeating pixels, and then boundary detection is performed. This can be done by detecting the boundary of the merged fixed pattern at high resolution based on the pixel value gradient change (step S113). Then, based on the boundary detection result, the first fixed pattern with higher resolution is eroded (step S115) to obtain a clear fixed pattern. This allows the system to perform motion estimation and motion compensation based on the detection result of the obtained clear fixed pattern (step S117). The goal is to obtain a dynamic image with a clear fixed pattern and improved frame rate after motion estimation and motion compensation processing.

[0057] According to one embodiment, in the operation process, in step S105 of detecting multiple fixed patterns in frame images with multi-level resolution, the frame images with the highest and lowest resolution can be discarded, that is, computing power is saved by discarding part of the steps of detecting fixed patterns.

[0058] It is worth mentioning that in the fixed pattern detection method in the aforementioned dynamic image, when acquiring frame images of multiple resolutions from the dynamic image, when performing erosion processing on merged fixed patterns at different (lower) resolutions, the erosion processing is further performed based on the boundary detection result after converting a lower-resolution merged fixed pattern obtained from the lower resolution level to a lower resolution. The relevant implementation process can be found in the following examples. Figure 11 The diagram shows an embodiment of a fixed pattern processing method between multiple resolution levels.

[0059] In the following examples, the image resolution values ​​such as 4K and 2K are only used to illustrate the implementation of the fixed pattern detection method in dynamic images, and are not intended to limit the scope of the invention.

[0060] When the system acquires a 4K dynamic image, it can simultaneously acquire images at multiple resolutions. This example shows that the multiple resolutions include 4K, 2K, 1K, 0.5K, and 0.25K. Fixed pattern detection can be performed on each resolution. For example, fixed pattern detection (111) is performed on the 4K image based on the feature differences between consecutive frames; fixed pattern detection (115) is performed on the 2K image; fixed pattern detection (122) is performed on the 1K image; fixed pattern detection (129) is performed on the 0.5K image; and fixed pattern detection (136) is performed on the 0.25K image. Furthermore, when performing erosion processing on the fixed patterns obtained at each resolution, the boundary detection results of the lower resolution image are also referenced.

[0061] According to the example shown in the figure, a fixed pattern detection is first performed on the 4K image (111), and the statistical information generated by the relevant results is used to form a historical record (114), which can be used as a confidence index for subsequent fixed pattern detection. At the same time, the 4K fixed pattern is converted to 2K resolution (112) and the result obtained from fixed pattern detection on the 2K image (115) is merged and calculated (116) to form a merged fixed pattern at 2K resolution (117).

[0062] The fixed pattern obtained by the fixed pattern detection (111) of the 4K image will be converted into a higher resolution (4K) image (119) by 2K erosion processing (118) at a lower resolution (i.e., 2K) and 4K fixed pattern boundary detection (120) will be performed. After the lower resolution (2K) edge reduction operation, the system can add back the higher resolution (4K) boundary according to the 4K boundary detection result, and then perform higher resolution (4K) erosion processing (113), that is, remove the redundant pixels around the fixed pattern.

[0063] Accordingly, when performing 2K erosion processing (118), the boundary detection results of the lower resolution fixed pattern are referenced before erosion processing.

[0064] Next, the 2K resolution merged fixed pattern (117) obtained in the above steps is converted to a lower resolution (1K) (121), and then merged with the fixed pattern obtained by performing fixed pattern detection (122) on the 1K resolution image (123) to obtain the merged fixed pattern at 1K resolution (124).

[0065] Next, the 1K resolution fixed pattern is subjected to erosion processing (125), including conversion to a higher resolution, i.e., 1K to 2K (126), and boundary detection of the 2K resolution fixed pattern (127). After edge reduction operation at the lower resolution (1K at this stage), the boundary of the higher resolution (2K at this stage) is added back based on the boundary detection result, so as to continue the erosion processing of the 2K resolution fixed pattern (118). Similarly, when performing the erosion processing of the 1K resolution fixed pattern (125), the boundary detection result obtained after processing the lower resolution fixed pattern is also referenced.

[0066] When the merged fixed pattern at 1K resolution is obtained (124), it is converted down to 0.5K (128), and merged with the result of the 0.5K fixed pattern detection (129) (130) to obtain the 0.5K merged fixed pattern (131). Similarly, the 0.5K resolution fixed pattern is eroded (132), and the 0.5K resolution is converted to a higher resolution 1K fixed pattern (133), and 1K boundary detection is performed (134). After the lower resolution (0.5K at this stage) edge reduction operation, the higher resolution (1K at this stage) boundary is added back according to the boundary detection result, and then the above 1K resolution fixed pattern erosion process is performed (125).

[0067] Similarly, the merged fixed pattern (131) obtained by 0.5K merging calculation is obtained, and then the fixed pattern (135) is converted down to 0.25K resolution. This fixed pattern is then merged with the fixed pattern (136) obtained by 0.25K fixed pattern detection (137) to obtain the 0.25K merged fixed pattern (138).

[0068] After the 0.25K merged fixed pattern is converted to a higher resolution (e.g., 0.5K) fixed pattern (139), and the 0.5K boundary detection is performed (140), similarly, the lower resolution (0.25K) edge reduction operation is performed first, and the higher resolution (0.5K) boundary is added back according to the boundary detection result. Then the system performs erosion processing on the 0.5K resolution fixed pattern (132).

[0069] Thus, to obtain a clear and fixed pattern after processing, the fixed pattern detection method in dynamic images proposed in this disclosure employs multi-level resolution calculations to obtain fixed patterns at different resolutions. Based on the results of the fixed patterns obtained at each resolution level, as well as the conversion calculations between different resolutions, and with reference to the boundary information detected by the lower resolution fixed pattern, a clear fixed pattern after erosion processing at a higher or highest resolution can finally be obtained.

[0070] In particular, the system sets a flag for each frame pixel of a clear fixed pattern, for example, setting the flag to "1". In addition to being stored in memory as a historical record for evaluating the confidence level of the clear fixed pattern, it also serves as the basis for performing motion estimation and motion compensation to avoid processing fixed patterns.

[0071] Furthermore, based on historical records, areas where fixed patterns frequently appear can be identified. For example, common logo patterns only appear in a few specific areas, such as the four ends of the screen. Therefore, fixed pattern detection can be performed only on these few areas, which can reduce the amount of computation.

[0072] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of the patent application of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention specification and drawings are included in the scope of the patent application of the present invention.

[0073] [Symbol Explanation]

[0074] 101: Frame N-1

[0075] 102: Frame N

[0076] 100: Object

[0077] 100': Interpolated object

[0078] 105: Compensated Image

[0079] 21: First Frame

[0080] 22: Second Frame

[0081] 200: Object

[0082] 210, 210': Logo design

[0083] 20: Interpolated Frames

[0084] 200': Interpolated object

[0085] 30, 30': Logo design

[0086] 501: Previous Frame Data

[0087] 502: Current frame data

[0088] 505: Fixed Pattern Detection Unit

[0089] 507: Fixed Pattern Logo

[0090] 509: Historical Records

[0091] 60: frames

[0092] 600: Fixed pattern

[0093] 701: Dynamic Image Data

[0094] 702: Buffer

[0095] 703: Clear and fixed pattern

[0096] 704: Motion-Compensated Frame Rate Conversion Unit

[0097] 705: Image data after frame rate conversion

[0098] 706: Statistical Information

[0099] 707: Motion Estimation Processing Unit

[0100] 708: Processing Unit

[0101] 801: Second-resolution fixed pattern detection

[0102] 803: Consolidated Calculation

[0103] 805: First resolution downgraded to second resolution

[0104] 807: Merge Fixed Patterns

[0105] 901: First resolution merging

[0106] 903: First-Resolution Erosion

[0107] 905: Second-Resolution Erosion

[0108] 907: Second resolution upgraded to first resolution

[0109] 909: Detecting the first resolution boundary

[0110] 111:4K Fixed Pattern Detection

[0111] 112: 4K to 2K

[0112] 113:4K Erosion Treatment

[0113] 114: Statistical Information

[0114] 115:2K Fixed Pattern Detection

[0115] 116: Combined Calculation

[0116] 117:2K merge

[0117] 118:2K erosion treatment

[0118] 119:2K to 4K

[0119] 120:4K Boundary Detection

[0120] 121:2K to 1K

[0121] 122:1K Fixed Pattern Detection

[0122] 123: Combined Calculation

[0123] 124:1K merge

[0124] 125:1K erosion treatment

[0125] 126:1K to 2K

[0126] 127:2K Boundary Detection

[0127] 128:1K to 0.5K

[0128] 129:0.5K Fixed Pattern Detection

[0129] 130: Combined Calculation

[0130] 131:0.5K merge

[0131] 132:0.5K erosion treatment

[0132] 133: 0.5K to 1K

[0133] 134:1K Boundary Detection

[0134] 135: 0.5K to 0.25K

[0135] 136:0.25K Fixed Pattern Detection

[0136] 137: Combined Calculation

[0137] 138:0.25K merge

[0138] 139: 0.25K to 0.5K

[0139] 140:0.5K Boundary Detection

[0140] 11: First Window

[0141] 12: Second Window

[0142] 13: Third Window

[0143] Steps S101 to S117 utilize the detection results of the fixed pattern in the dynamic image to perform MEMC.

Claims

1. A method for detecting a fixed pattern in a dynamic image, executed in a system, comprising: Acquire continuous dynamic images; Extract multi-resolution frame images from the dynamic image; Based on the features of the preceding and following frames, multiple fixed patterns are detected in the multi-resolution frame images, including at least a first fixed pattern with higher resolution and a second fixed pattern with lower resolution. The first fixed pattern with a higher resolution is converted into a converted fixed pattern with a lower resolution, and the converted fixed pattern and the second fixed pattern are merged at the same lower resolution to obtain a merged fixed pattern. The merged fixed pattern at a lower resolution is subjected to erosion processing; The merged fixed pattern after erosion treatment is converted into a higher resolution pattern, and boundary detection is performed; and Based on the results of boundary detection, the first fixed pattern with higher resolution is subjected to erosion processing to obtain a clear fixed pattern.

2. The method for detecting fixed patterns in dynamic images as described in claim 1, wherein, In the step of detecting multiple fixed patterns in the multi-resolution frame image, a matrix scanning window is used to scan the pixel values ​​of each frame image, and the fixed patterns at each resolution are detected based on the absolute difference between the pixel values ​​of consecutive frames.

3. The method for detecting fixed patterns in dynamic images as described in claim 1, wherein, The merged fixed pattern after erosion is converted into a higher resolution fixed pattern by repeating pixels, and then the boundary of the merged fixed pattern at high resolution is detected based on the pixel value gradient change in the fixed pattern.

4. The method for detecting fixed patterns in dynamic images as described in claim 1, wherein, In the frame images with multiple resolutions obtained from the dynamic image, when erosion processing is performed on the merged fixed pattern at a lower resolution, the erosion processing is further performed based on the boundary detection results after the lower resolution merged fixed pattern obtained from the lower resolution level is converted to a lower resolution.

5. The method for detecting fixed patterns in dynamic images as described in claim 1, wherein, After the eroded and merged fixed pattern is converted into a higher resolution pattern, and after boundary detection, i.e. after lower resolution edge reduction operation and re-adding back the higher resolution boundary, the first fixed pattern with higher resolution is eroded again.

6. The method for detecting fixed patterns in dynamic images as described in any one of claims 1 to 5, wherein, After obtaining the clear fixed pattern, motion estimation and motion compensation are performed on the dynamic image based on the clear fixed pattern to improve the frame rate of the dynamic image.

7. The method for detecting fixed patterns in dynamic images as described in claim 6, wherein, Once the clear fixed pattern is obtained, the pixels of each frame that are determined to be the clear fixed pattern are set as flag "1" and recorded in a memory. When motion estimation and motion compensation are performed on the dynamic image, the pixels that have been set as flag "1" are ignored based on the memory record.

8. The method for detecting fixed patterns in dynamic images as described in claim 7, wherein, In setting the markers for each frame pixel of the sharp, fixed pattern, the confidence level of the sharp, fixed pattern is further evaluated by referring to the historical records stored in the memory.

9. A system for performing fixed pattern detection, comprising: A processing circuit executes a fixed pattern detection method in a dynamic image via firmware, including: Acquire continuous dynamic images; Extract multi-resolution frame images from the dynamic image; Based on the features of the preceding and following frames, multiple fixed patterns are detected in the multi-resolution frame images, including at least a first fixed pattern with higher resolution and a second fixed pattern with lower resolution. The first fixed pattern with a higher resolution is converted into a converted fixed pattern with a lower resolution, and the converted fixed pattern and the second fixed pattern are merged at the same lower resolution to obtain a merged fixed pattern. The merged fixed pattern at a lower resolution is subjected to erosion processing; The merged fixed pattern after erosion treatment is converted into a higher resolution pattern, and boundary detection is performed; and Based on the results of boundary detection, the first fixed pattern with higher resolution is subjected to erosion processing to obtain a clear fixed pattern.

10. The system of claim 9, wherein, After obtaining the clear fixed pattern, motion estimation and motion compensation are performed on the dynamic image based on the clear fixed pattern to improve the frame rate of the dynamic image. When the clear fixed pattern is obtained, the pixels of each frame that are determined to be the clear fixed pattern are set as flag "1" and recorded in a memory. When motion estimation and motion compensation are performed on the dynamic image, the pixels that have been set as flag "1" are ignored based on the memory record.

Images