Image adjustment method and image adjustment device
Patent Information
- Application Number
- CN202610396998.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2025-04-10
- Filing Date
- 2026-03-30
- Publication Date
- 2026-10-20
AI Technical Summary
在非单一光源的场景或复杂场景的状况下，若采用上述传统统计方法，将无法兼顾场景内不同区域的亮度和颜色效果
[0007]本申请的实施例所体现的技术手段可以改善先前技术的缺点的至少其中之一。本申请的影像调整方法及影像调整装置可采用空间透视影像增强算法以立体化前景区域(如人像)，并远景化背景区域，以使前景区域(如人像)于视觉上更为立体与生动。再者，本申请的影像调整方法及影像调整装置可采用深度统计影像增强算法以防止单一统计值应用到输入影像的整张画面，造成画面偏差的问题。
Smart Images
Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] This application relates to image adjustment methods and image adjustment apparatus, and more particularly to image adjustment methods and image adjustment apparatus for optimizing output images. Background Technology
[0002] Traditional statistical methods for image quality adjustment are mostly two-dimensional, such as global statistical values and face statistical values. In scenes with multiple light sources or complex scenes, these traditional statistical methods cannot simultaneously take into account the brightness and color effects of different areas within the scene.
[0003] Taking color as an example, users often adjust skin tones in images to appear more rosy. However, adjusting skin tones in complex lighting scenarios can actually lead to color cast issues. For instance, in backlit scenes where the light source includes sunlight from outside the window and indoor lighting, adjusting the image based on skin tone will result in color casts in the sky and trees outside the window. Summary of the Invention
[0004] In view of the shortcomings of the prior art, one of the objectives of this application is (but not limited to) to provide an image adjustment method and an image adjustment device to improve the shortcomings of the prior art.
[0005] In some embodiments, the processor executes an image adjustment method according to at least one instruction. The image adjustment method includes: providing a depth map; performing a spatial perspective image enhancement algorithm on the depth map to generate a spatial perspective enhanced image; performing foreground and background image statistics on the input image and the depth map to generate foreground image statistics and background image statistics; and performing a depth statistics image enhancement algorithm on the spatial perspective enhanced image, the foreground image statistics, and the background image statistics to generate an output image.
[0006] In some embodiments, the image adjustment apparatus includes a processor. The processor performs the following steps according to at least one instruction: providing a depth map; performing a spatial perspective image enhancement algorithm on the depth map to generate a spatial perspective enhanced image; performing foreground and background image statistics on the input image and the depth map to generate foreground image statistics and background image statistics; and performing a depth statistics image enhancement algorithm on the spatial perspective enhanced image, the foreground image statistics, and the background image statistics to generate an output image.
[0007] The technical means embodied in the embodiments of this application can improve at least one of the shortcomings of prior art. The image adjustment method and image adjustment apparatus of this application can employ a spatial perspective image enhancement algorithm to stereoscopically enhance the foreground area (such as a human portrait) and exaggerate the background area, so that the foreground area (such as a human portrait) appears more three-dimensional and vivid. Furthermore, the image adjustment method and image adjustment apparatus of this application can employ a depth statistical image enhancement algorithm to prevent a single statistical value from being applied to the entire image of the input image, thus avoiding image deviation.
[0008] The features, implementation, and effects of this application are described in detail below with reference to the accompanying drawings, showing preferred embodiments. Attached Figure Description
[0009] Figure 1 A schematic diagram of an image adjustment device is provided according to some embodiments of this application; Figure 2 A flowchart of an image adjustment method is provided based on some embodiments of this application. Figure 3 A flowchart for generating a depth map is provided according to some embodiments of this application; Figure 4 A flowchart for generating spatial perspective enhanced images is provided based on some embodiments of this application. Figure 5A A schematic diagram illustrating an input image according to some embodiments of this application; Figure 5B A schematic diagram illustrating a spatial perspective enhancement image according to some embodiments of this application; Figure 6A A schematic diagram illustrating an input image according to some embodiments of this application; Figure 6B A schematic diagram illustrating a spatial perspective enhancement image according to some embodiments of this application; Figure 7A A schematic diagram illustrating an input image according to some embodiments of this application; Figure 7B A schematic diagram illustrating a spatial perspective enhancement image according to some embodiments of this application; Figure 8 A flowchart for generating foreground and background image statistics is provided based on some embodiments of this application. Figure 9A A schematic diagram illustrating a depth map according to some embodiments of this application; Figure 9B A schematic diagram illustrating a custom depth image based on some embodiments of this application; Figure 10A A schematic diagram illustrating a depth map according to some embodiments of this application; Figure 10B A schematic diagram illustrating a custom depth image based on some embodiments of this application; Figure 11 A flowchart for generating an output image is provided according to some embodiments of this application; Figure 12A A schematic diagram illustrating an input image according to some embodiments of this application; Figure 12B A schematic diagram illustrating a depth statistical enhancement image according to some embodiments of this application; Figure 13A A schematic diagram illustrating an input image according to some embodiments of this application; Figure 13B A schematic diagram illustrating an output image of automatic exposure control according to some embodiments of this application; Figure 14A A schematic diagram illustrating an input image according to some embodiments of this application; and Figure 14B This is a schematic diagram illustrating an output image of automatic exposure control according to some embodiments of this application. Detailed Implementation
[0010] All terms used herein have their common meanings. The definitions of the terms used above in commonly used dictionaries, and any examples of the use of any term discussed herein, are merely illustrative and should not be construed as limiting the scope or meaning of this application. Similarly, this application is not limited to the various embodiments shown in this specification.
[0011] As used herein, "coupling" or "connection" can refer to two or more components making direct physical or electrical contact with each other, or indirectly making direct physical or electrical contact with each other, or to two or more components operating or moving together. As used herein, the term "circuit" can refer to a device in which at least one transistor and / or at least one active or passive component are connected in a certain manner to process signals.
[0012] As used herein, the term "and / or" includes any combination of one or more of the listed related items. The terms "first," "second," and "third," etc., are used herein to describe and distinguish the individual components. Therefore, a first component may also be referred to as a second component without departing from the intent of this application. For ease of understanding, similar components in the various figures will be designated with the same reference numerals.
[0013] Figure 1A schematic diagram of an image adjustment apparatus 100 is provided according to some embodiments of this application. As shown, the image adjustment apparatus 100 includes a processor 110 and a memory 120. The memory 120 is used to store at least one instruction. The processor 110 is used to read at least one instruction to execute an image adjustment process. For ease of understanding of the operation of the image adjustment apparatus 100, please also refer to... Figure 2 , Figure 2 This is a flowchart based on some image adjustment methods 200.
[0014] Please see Figure 2 Step 210 provides a depth map. For example, see [link to example]. Figure 3 Step 310 involves providing an input image, such as the input image for the current frame. See also... Figure 3 Step 320 provides the historical input image. See [link / reference]. Figure 3 Step 330 provides a historical depth map. See also... Figure 3 Step 340 involves reducing the input image to produce a reduced image. For example, this application may use bilinear interpolation to reduce the input image, and the width and height of the reduced image may be 256. 160. However, this application is not limited to the above embodiments. In other embodiments, other suitable techniques may be used to reduce the size, and other suitable values may be used for the width and height of the reduced image, depending on the actual needs.
[0015] Please see Figure 3 Step 350 involves performing depth estimation on the reduced image to generate a depth map. For example, this application may employ a neural network-like system to perform depth estimation, and the depth value of the depth map output by the neural network-like system can be 0 to FLOAT_MAX. See also... Figure 3 In step 360, this application can normalize the depth map to make the depth map lie within a numerical range. For example, this application can use a depth estimation neural network to perform normalization, as shown in the following formula: …Formula 1 Please refer to Formula 1. The depth map output by the neural network can have a depth value ranging from 0 to FLOAT_MAX, which is then normalized using Formula 1 to be between 0 and 255. However, this application is not limited to the above embodiments. In other embodiments, the normalized value can be any other suitable value, depending on the actual needs.
[0016] Please see Figure 3Step 370 involves performing a temporal smoothing algorithm on the depth map based on the historical input image and the historical depth map. The depth map is then output in step 380. For example, the depth estimation neural network of this application employs single-frame prediction and executes a temporal smoothing algorithm between frames. The input to the temporal smoothing algorithm is the input image of the current frame, which is then combined with the historical input image and the historical depth map to calculate the pixel value differences between frames (e.g., RGB or YUV color gamut), thereby defining dynamic and static regions. Subsequently, this application performs depth map value averaging in the static region, while updating the depth map of the current frame based on the pixel value differences in the dynamic region. It should be noted that if the image adjustment device 100 is applied to a depth camera composed of two lenses, the depth map generated by the depth camera can be directly used, without the aforementioned depth map generation steps 310-380, and combined with the subsequent image enhancement algorithms of this application.
[0017] Please refer back to this. Figure 2 In step 220, a spatial perspective image enhancement algorithm is applied to the depth map to generate a spatial perspective enhanced image. For example, please refer to... Figure 4 Step 410 involves providing an input image. Please refer to [link / reference]. Figure 4 Step 420 provides a depth map. See [link / reference] Figure 4 Step 430 involves calculating the weights based on the depth map, using the following formula: …Formula 2 …Formula 3 As shown in formulas 2 and 3 As weight, It is a conversion type. For depth map, and The parameters are defined as follows. For example, the foreground has a higher weight, while the background has a lower weight. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0018] Please see Figure 4 Step 440 involves generating luminance contrast based on weights, the local luminance map of the input image, and luminance and contrast enhancement curves, using the following formula: …Formula 4 As shown in Formula 4, For brightness contrast, As weight, This is a local brightness map. Input brightness, The above-described embodiments are for brightness and contrast enhancement. However, this application is not limited to these embodiments, but is merely used to illustrate one implementation method of this application.
[0019] Please see Figure 4 Step 450 involves generating color saturation based on weights, saturation scaling factors, and the color components of the input image, using the following formula: …Formula 5 As shown in Formula 5, For color saturation, As weight, This is the saturation scaling factor. The color components are the input image. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0020] Please see Figure 4 Step 460 generates detail strength control coefficients based on brightness contrast, weight, texture scaling factor, and brightness difference of the input image, using the following formula: …Formula 6 As shown in Formula 6, For the control coefficient of detail strength, For brightness contrast, As weight, This is the texture scaling factor. The brightness difference of the input image. Input brightness, This is a local brightness map. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0021] In some embodiments, please refer to Figure 4 In step 490, this application can generate spatial perspective enhanced images based on brightness contrast, color saturation, and detail strength control coefficients. For example, based on the principle of spatial perspective (or atmospheric perspective), the foreground area closer to the user is the area with the strongest visual impact, with the location of the human figure being the primary focus. To enhance the three-dimensionality of the foreground area (such as the human figure), the spatial perspective image enhancement algorithm references the depth map and enhances multiple image parameters of the foreground area. These image parameters can be brightness contrast, color saturation, detail strength control coefficients (corresponding to detail richness), etc. Conversely, multiple image parameters of the background area can be reduced to achieve a more distant spatial perspective effect. These image parameters can be brightness contrast, color saturation, detail strength control coefficients (corresponding to detail richness), etc.
[0022] Figure 5A A schematic diagram of an input image is provided according to some embodiments of this application. Figure 5BThis is a schematic diagram illustrating a spatial perspective enhancement image according to some embodiments of this application. Please refer to... Figure 5A , 5B , Figure 5B Adjusted Figure 5A The brightness contrast of the foreground area (such as a human figure) in the input image is determined by... Figure 5B It can be seen that after improving the brightness and contrast of the foreground area (such as a portrait), the portrait appears more three-dimensional and vivid, and the portrait will not be too dark due to backlighting.
[0023] Figure 6A A schematic diagram of an input image is provided according to some embodiments of this application. Figure 6B This is a schematic diagram illustrating a spatial perspective enhancement image according to some embodiments of this application. Please refer to... Figure 6A , 6B , Figure 6B Adjusted Figure 6A The color saturation of the foreground area (such as a human figure) in the input image is determined by... Figure 6B It can be seen that by increasing the color saturation of the foreground area (such as a portrait), the portrait appears more three-dimensional and vivid, allowing the user's vision to focus more on the portrait.
[0024] Figure 7A A schematic diagram of an input image is provided according to some embodiments of this application. Figure 7B This is a schematic diagram illustrating a spatial perspective enhancement image according to some embodiments of this application. Please refer to... Figure 7A , 7B , Figure 7B Adjusted Figure 7A The detail strength control coefficient of the foreground area (such as a portrait) in the input image is determined by... Figure 7B It can be seen that after enhancing the details of the foreground area (such as a portrait), the portrait appears more three-dimensional and vivid.
[0025] Please see Figure 4 In step 470, this application can further generate a preference color coefficient based on the weight, style color offset scaling factor, and color components of the input image, as shown in the following formula: …Formula 7 As shown in Formula 7, For color preference coefficient, As weight, This is the scaling factor for style color offset. The color components are the input image. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0026] Please see Figure 4In step 480, this application can further generate noise reduction strength control coefficients based on the maximum depth value of the depth map, weights, strong noise reduction color space parameters, and weak noise reduction color space parameters, as shown in the following formula:
[0027] …Formula 8 As shown in Formula 8, This is the noise reduction strength control coefficient. This is the maximum depth value. As weight, For color space parameters, To perform strong noise reduction on the color space parameters to produce strongly noise-reduced color space parameters. To generate weakly denoised color space parameters, weak noise reduction is performed on the color space parameters. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0028] In some embodiments, please refer to Figure 4 In step 490, this application can also generate a spatial perspective enhanced image based on brightness contrast, color saturation, detail strength control coefficient, the aforementioned color preference coefficient, and the aforementioned noise reduction strength control coefficient. To enhance the three-dimensionality of the foreground area (such as a portrait), the spatial perspective image enhancement algorithm references the depth map and improves multiple image parameters of the foreground area. These image parameters include brightness contrast, color saturation, detail strength control coefficient (corresponding to detail richness), the aforementioned color preference coefficient, and the aforementioned noise reduction strength control coefficient. Conversely, multiple image parameters of the background area can be reduced to achieve a more distant perspective effect. These image parameters also include brightness contrast, color saturation, detail strength control coefficient (corresponding to detail richness), the aforementioned color preference coefficient, and the aforementioned noise reduction strength control coefficient.
[0029] Please refer back to this. Figure 2 In step 230, foreground and background image statistics are performed on the input image and depth map to generate foreground image statistics and background image statistics. For example, please refer to... Figure 8 Step 810 involves providing an input image. Please refer to [link / reference]. Figure 8 Step 820 provides a depth map. See also... Figure 8 Step 830 provides the Region of Interest (ROI) for the face.
[0030] Please see Figure 8 Step 840 involves setting a foreground depth threshold based on the facial region of interest. For example, this application can obtain the facial region of interest from the input image. Subsequently, the average depth value of the facial region of interest is calculated as the foreground depth threshold.
[0031] Please see Figure 8 Step 850: Based on the foreground depth threshold, the input image and the depth map are distinguished to identify the foreground and background regions, as shown in the following formula: …Formula 9 …Formula 10 As shown in formulas 9 and 10 For depth map, Foreground depth threshold For a custom depth image, if the pixel value of the depth map is greater than the foreground depth threshold, the pixel value is set to 255, i.e., set to white; if the pixel value of the depth map is less than or equal to the foreground depth threshold, the pixel value is set to 0, i.e., set to black, to generate the custom depth image. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0032] Figure 9A This is a schematic diagram illustrating a depth map according to some embodiments of this application. Figure 9B This is a schematic diagram illustrating a custom depth image based on some embodiments of this application. Please refer to... Figure 9A , 9B According to the foreground depth threshold Figure 9A The depth map was redefined to generate Figure 9B Custom depth imagery. (By...) Figure 9B It can be seen that the distinction between the foreground region 910 and the background region 920 is more significant.
[0033] Figure 10A This is a schematic diagram illustrating a depth map according to some embodiments of this application. Figure 10B This is a schematic diagram illustrating a custom depth image based on some embodiments of this application. Please refer to... Figure 10A , 10B Based on the foreground depth threshold and the region of interest of the face, Figure 10A The depth map was redefined to generate Figure 10B Custom depth imagery. (By...) Figure 10B It can be seen that it only distinguishes the foreground part corresponding to the region of interest of the face, making the distinction between the foreground region 1010 and the background region 1020 corresponding to the face more significant.
[0034] Please see Figure 8 Steps 860 and 870 involve performing foreground and background image statistics on the foreground and background regions to generate foreground image statistical values and background image statistical values, using the following formulas: …Formula 11 …Formula 12 …Formula 13 …Formula 14 …Formula 15 As shown in formulas 11-15, For custom depth images, The input image is the image itself. If the pixel value of the custom-defined depth image is 255, it is counted as a foreground image statistical value. If the pixel value of the custom-defined depth image is 0, it is counted as a background image statistical value. In some embodiments, this application can calculate a foreground histogram of the foreground region. Using the statistical values of the foreground image as a reference, and calculating the background histogram of the background region. This is used as a statistical value for the background image. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0035] Please refer back to this. Figure 2 In step 240, a depth statistical image enhancement algorithm is applied to the spatial perspective enhanced image, foreground image statistics, and background image statistics to generate an output image. For example, please refer to... Figure 11 Step 1110 involves providing an input image. Please refer to [link / reference]. Figure 11 Step 1120 provides a depth map. See [link / reference]. Figure 11 Step 1130 provides statistical values for the foreground and background images. Please refer to [link / reference]. Figure 11 Step 1140: Execute the brightness adjustment algorithm, the formula is as follows:
[0036] …Formula 16
[0037] …Formula 17 As shown in formulas 16 and 17 Background variability of background image statistics. Foreground variability is the statistical value of the foreground image. This represents the background average value of the background image statistics. This represents the foreground average value of the foreground image statistics. As shown in Formula 16, if the background variability of the background image statistics is much smaller than the foreground variability of the foreground image statistics, and the background average value of the background image statistics is greater than the foreground average value of the foreground image statistics, the automatic exposure target brightness of the input image should be reduced. To enhance its dynamic range ( As shown in Formula 17, if the background variability of the background image statistics is much greater than the foreground variability of the foreground image statistics, and the background average value of the background image statistics is greater than the foreground average value of the foreground image statistics, the automatic exposure target brightness of the input image is enhanced. To enhance its dynamic range ( However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0038] Please see Figure 11 Step 1150: Execute the color temperature adjustment algorithm, the formula is as follows:
[0039] …Formula 18
[0040] …Formula 19 As shown in formulas 18 and 19 For foreground automatic white balance gain, Automatic white balance gain for background. Adjust parameters for foreground color. This is used to adjust the background color parameters. As shown in Formula 18, the foreground automatic white balance gain is calculated based on the foreground image statistics, and this foreground automatic white balance gain is then used for the foreground color adjustment parameters. As shown in Formula 19, based on the background image statistics (... The background automatic white balance gain is calculated and used as a background color adjustment parameter. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0041] …Formula 20 …Formula 21 …Formula 22 As shown in formulas 20-22, Adjust parameters for foreground color. Adjust the parameters for the background color. The input image contains RGB values. To output images, Foreground weights of the depth map, This represents the background weight of the depth map. As shown in Equations 20 and 21, a white balance algorithm is performed on the RGB values of the input image to generate foreground color adjustment parameters. and background color adjustment parameters As shown in Formula 22, this application can determine the foreground weights based on the depth map. Foreground color adjustment parameters Background weights of depth maps and background color adjustment parameters To adjust the output image However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0042] Please see Figure 11 Step 1160 involves executing a facial skin tone correction algorithm. For example, this application can perform the facial skin tone correction algorithm based on the facial region of interest provided in step 1170. Please refer to... Figure 11 Step 1180 provides the output image.
[0043] Figure 12A A schematic diagram of an input image is provided according to some embodiments of this application. Figure 12B This is a schematic diagram illustrating a depth statistical enhancement image according to some embodiments of this application. Please refer to... Figure 12A , 12B ,right Figure 12A The input image is subjected to a depth statistical image enhancement algorithm to produce Figure 12B This involves depth statistical image enhancement. Besides spatial perspective image enhancement algorithms, depth statistical image enhancement algorithms utilize statistical values from the foreground and background images to enhance different depth maps separately. It also strengthens the guidance of statistics to prevent the application of a single statistical value to the entire input image, thus avoiding partial or overall image distortion. Taking a backlit scene as an example, the area outside the window, which is farther away, is a bright area, while the indoor figures, which are closer in depth, are dark areas. Through statistical guidance of the depth map, such as Auto Exposure (AE) and Auto White Balance (AWB) strategies, the brightness of the dark indoor areas and the brightness of the bright areas outside the window can be enhanced separately, correcting the previous situation where the high brightening ratio of dark areas suppressed the bright areas in image processing. When the scene has complex mixed light sources, this application can also optimize and adjust the white balance of different depth regions through depth region division.
[0044] In some embodiments, using a depth map in conjunction with a region of interest (ROI) for facial contouring can enhance stable 3A control (such as auto exposure (AE) control, auto white balance (AWB) control, and auto focus (AF) control) or adjust skin tone. The following section will use auto exposure (AE) control as an example; please refer to [link to relevant documentation]. Figure 13A , 13B , Figure 13A A face was detected; however, when Figure 13B If no face is detected, it will cause instability in the automatic exposure (AE) control, such as... Figure 13B The brightness in the image drops significantly. In addition, face detection may fail to detect faces or become unstable due to different light sources, different skin tones, complex backgrounds, or changes in the angle of the face. Both failure to detect faces and unstable detection will lead to instability in automatic exposure (AE) control.
[0045] To address the aforementioned issues, this application can refer to depth maps and historical face records to track face positions, thereby assisting in relevant judgments and maintaining consistent automatic exposure (AE) control settings (such as maintaining the same exposure strategy), thus optimizing the stability of AE control. This application can employ a 3A stabilization algorithm, the formula of which is as follows:
[0046]
[0047] …Formula 23 As shown in Formula 23 The current depth map of the region of interest for a face. Historical depth map of regions of interest for faces. This is the depth threshold. As shown in Formula 23, if the current depth map of the region of interest for the face... Historical depth map of facial regions of interest The difference is less than the depth threshold The system maintains the automatic exposure parameters, automatic white balance parameters, automatic focus parameters, brightness, color temperature, and sharpness of the input image. However, this application is not limited to the above embodiments, which are merely illustrative of one implementation method of this application.
[0048] Figure 14A A schematic diagram of an input image is provided according to some embodiments of this application. Figure 14B This is a schematic diagram illustrating an output image of automatic exposure control according to some embodiments of this application. For example... Figure 14A , 14B As shown, due to the 3A stabilization algorithm of this application, the brightness of the two images remains stable.
[0049] It should be noted that this application is not limited to the embodiments shown in Figures 1 to 14B, but is merely used to illustrate one implementation of this application to facilitate understanding of the technology. The patent scope of this application shall be determined by the scope of the invention application. Modifications and refinements made by those skilled in the art to the embodiments of this application without departing from the spirit of this application still fall within the scope of the invention application.
[0050] The technical means embodied in the embodiments of this application can improve at least one of the shortcomings of prior art. The image adjustment method and image adjustment apparatus of this application can employ a spatial perspective image enhancement algorithm to stereoscopically enhance the foreground area (such as a human portrait) and exaggerate the background area, so that the foreground area (such as a human portrait) appears more three-dimensional and vivid. Furthermore, the image adjustment method and image adjustment apparatus of this application can employ a depth statistical image enhancement algorithm to prevent a single statistical value from being applied to the entire image of the input image, thus avoiding image deviation.
[0051] Although the embodiments of this application are described above, these embodiments are not intended to limit this application. Those skilled in the art can make changes to the technical features of this application based on the express or implied content of this application. All such changes may fall within the scope of patent protection sought by this application. In other words, the scope of patent protection of this application shall be determined by the claims of this specification.
[0052] [Symbol Explanation] 100: Image adjustment device 110: Processor 120: Memory 200: Method 210~240: Steps 300: Method 310~380: Steps 400: Method 410~490: Steps 800: Method 810~870: Steps 910: Foreground Area 920: Background Area 1010: Foreground Area 1020: Background Area 1100: Method 1110~1180: Steps
Claims
1. An image adjustment method, wherein a processor executes the image adjustment method according to at least one instruction, characterized in that, Include: Provide a depth map; A spatial perspective image enhancement algorithm is applied to the depth map to generate a spatial perspective enhanced image. Perform foreground and background image statistics on an input image and the depth map to generate a foreground image statistical value and a background image statistical value; and A depth statistical image enhancement algorithm is applied to the spatial perspective enhanced image, the foreground image statistics, and the background image statistics to generate an output image.
2. The image adjustment method as described in claim 1, wherein applying the spatial perspective image enhancement algorithm to the depth map to generate the spatial perspective enhanced image comprises: Calculate a weight based on the depth map; A brightness contrast is generated based on the weight, a local brightness map of the input image, and a brightness and contrast enhancement curve. A color saturation is generated based on the weight, a saturation scaling factor, and a color component of the input image. A detail strength control coefficient is generated based on the brightness contrast, the weight, a texture scaling factor, and a brightness difference of the input image. as well as The spatial perspective enhancement image is generated based on the brightness contrast, the color saturation, and the detail strength control coefficient.
3. The image adjustment method as described in claim 2, wherein applying the spatial perspective image enhancement algorithm to the depth map to generate the spatial perspective enhanced image further comprises: A preference color coefficient is generated based on the weight, the same color offset scaling factor, and the color component of the input image. A noise reduction strength control coefficient is generated based on a maximum depth value of the depth map, the weight, a strong noise reduction color space parameter, and a weak noise reduction color space parameter; and The spatial perspective enhancement image is generated based on the brightness contrast, the color saturation, the detail strength control coefficient, the color preference coefficient, and the noise reduction strength control coefficient.
4. The image adjustment method as described in claim 1, wherein performing foreground and background image statistics on the input image and the depth map to generate the foreground image statistical value and the background image statistical value comprises: A foreground depth threshold is set based on the region of interest of a face; Based on the foreground depth threshold, the input image and the depth map are distinguished into a foreground region and a background region; as well as Perform foreground and background image statistics on the foreground and background regions to generate foreground image statistics and background image statistics.
5. The image adjustment method as described in claim 4, wherein setting the foreground depth threshold based on the region of interest of the face comprises: Obtain the region of interest of the face in the input image; and Calculate an average depth value for the region of interest of the face as a foreground depth threshold.
6. The image adjustment method of claim 5, wherein performing foreground and background image statistics on the foreground region and the background region to generate the foreground image statistical value and the background image statistical value comprises: Calculate a foreground histogram of the foreground region as a statistical value of the foreground image; and Calculate a background histogram for the background region as a statistical value for the background image.
7. The image adjustment method of claim 1, wherein performing the depth statistical image enhancement algorithm based on the spatial perspective enhanced image, the foreground image statistics, and the background image statistics to generate the output image comprises: If the background variability of the background image statistics is less than the foreground variability of the foreground image statistics, and the background average value of the background image statistics is greater than the foreground average value of the foreground image statistics, reduce the automatic exposure target brightness of the input image to enhance a dynamic range; or If the background variability of the background image statistics is greater than the foreground variability of the foreground image statistics, and the background average value of the background image statistics is greater than the foreground average value of the foreground image statistics, the brightness of the automatic exposure target of the input image is increased to enhance the dynamic range.
8. The image adjustment method as described in claim 1, wherein performing the depth statistical image enhancement algorithm based on the spatial perspective enhanced image, the foreground image statistics, and the background image statistics to generate the output image comprises: Calculate a foreground automatic white balance gain based on the statistical values of the foreground image, and use the foreground automatic white balance gain as a foreground color adjustment parameter; Calculate a background auto white balance gain based on the background image statistics, and use this background auto white balance gain for a background color adjustment parameter; and The output image is adjusted based on a foreground weight of the depth map, a foreground color adjustment parameter, a background weight of the depth map, and a background color adjustment parameter.
9. The image adjustment method as described in claim 1, further comprising: If the difference between a current depth map of a region of interest (ROI) and a historical depth map of the same ROI is less than a depth threshold, the automatic exposure parameters, automatic white balance parameters, automatic focus parameters, brightness, color temperature, and sharpness of the input image are maintained.
10. An image adjustment device, characterized in that, Include: A processor for performing the following steps according to at least one instruction: Provide a depth map; A spatial perspective image enhancement algorithm is applied to the depth map to generate a spatial perspective enhanced image. Perform foreground and background image statistics on an input image and the depth map to generate a foreground image statistical value and a background image statistical value; and A depth statistical image enhancement algorithm is applied to the spatial perspective enhanced image, the foreground image statistics, and the background image statistics to generate an output image.