A method for improving the definition of a projected image and a projector

Abstract

The application provides a method for improving the definition of a projector and a projector. The method comprises the following steps: acquiring an input projection image and a blank reference frame of a projection screen, constructing an illumination decay mask based on the blank reference frame, and correcting the input projection image using the illumination decay mask to obtain a corrected image. The corrected image is converted into a YUV color space image to separate a luminance component image and a chrominance component image. Based on the edge pixel proportion, smooth area proportion and local variance mean in the luminance component image, an image content complexity parameter is determined. Based on the complexity parameter, the luminance component image is sequentially subjected to bilateral filtering, Canny edge detection, edge superposition, bicubic interpolation and Laplacian sharpening to obtain an output image with improved definition. The definition is adaptively adjusted through the above method.

Description

Technical Field

[0001] This application relates to the field of projectors, and more particularly to a method for improving the image quality and clarity of a projector, as well as a projector in general. Background Technology

[0002] With the widespread application of projectors in scenarios such as conference displays, educational presentations, home theaters, and portable smart projectors, users' demands for the clarity, detail, and visual comfort of projected images are constantly increasing. Compared with ordinary flat panel display devices, the projector imaging process is also affected by various factors such as the optical system, projection distance, screen reflection characteristics, ambient light interference, and image compression transmission. These factors can easily lead to problems such as blurred edges, unclear text outlines, loss of detail, and uneven brightness distribution in local areas, thus affecting the viewing experience and information recognition.

[0003] Existing technologies typically employ image processing techniques such as sharpening filtering, contrast enhancement, edge enhancement, and resolution improvement to improve projected image quality. For example, this might involve filtering and denoising the input image before sharpening, enhancing contour information through edge detection and image overlay, or increasing display resolution through interpolation. While these solutions can improve image clarity to some extent, they are mostly designed for general image enhancement scenarios, often using fixed parameters or a single enhancement strategy. They rarely address specific issues arising during projector imaging, such as bright centers and dark edges, screen texture interference, and variations in the complexity of different image content.

[0004] However, existing projectors commonly suffer from the following problems in practical use: On the one hand, projector imaging is affected by factors such as the optical system, projection distance, screen reflection characteristics, ambient light, and image compression transmission, easily resulting in phenomena such as bright centers, dark edges, uneven brightness distribution, blurred edges, unclear text outlines, and loss of detail. On the other hand, existing technologies typically improve image quality through simple sharpening filters, contrast enhancement, or edge enhancement with fixed parameters. While these methods can improve local sharpness to some extent, they often suffer from drawbacks such as simultaneous amplification of noise, excessive edge enhancement, the production of white borders or ringing artifacts, and difficulty in adapting to different content scenarios. Therefore, a method and a projector that adaptively adjusts sharpness to improve projector image quality are needed. Summary of the Invention

[0005] In view of this, it is necessary to provide a method and a projector that adaptively adjusts the sharpness to improve the image quality of the projector, in order to solve the above problems.

[0006] Embodiments of this application provide a method for improving the image quality and clarity of a projector, comprising the following steps: S1. Obtain the input projection image and the blank reference frame of the projection screen, construct an illumination attenuation mask based on the blank reference frame, and use the illumination attenuation mask to perform illumination uniformity correction on the input projection image to obtain a corrected image. S2. Convert the corrected image into a YUV color space image to separate the luminance component image and the chrominance component image; S3. Determine the image content complexity parameters based on the proportion of edge pixels, the proportion of smooth regions, and the mean of local variance in the brightness component image. S4. Based on the complexity parameter, the brightness component image is sequentially subjected to bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening to obtain an output image with improved clarity.

[0007] In at least one embodiment of this application, the steps of sequentially performing bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening on the luminance component based on the complexity parameter to obtain an output image with improved sharpness include: Based on the content complexity parameter, the luminance component image is subjected to bilateral filtering to obtain a denoised luminance image; Canny edge detection is performed on the denoised brightness image to obtain an edge feature map; The edge feature map is superimposed on the deluminance component image to obtain the edge-enhanced image; Bicubic interpolation is performed on the edge enhancement image to obtain a high-resolution enhancement image; The high-resolution enhanced image is then subjected to Laplacian sharpening to obtain an output image with improved clarity.

[0008] In at least one embodiment of this application, the step of constructing an illumination attenuation mask based on the blank reference frame and using the illumination attenuation mask to perform illumination uniformity correction on the input projected image includes the following steps: The brightness distributions of the R, G, and B channels of the blank reference frame are extracted respectively to generate corresponding R channel attenuation masks, G channel attenuation masks, and B channel attenuation masks; The input projected image is normalized and corrected by the ratio of the original pixel value of each channel to the corresponding channel attenuation mask.

[0009] In at least one embodiment of this application, determining the image content complexity parameter based on the proportion of edge pixels, the proportion of smooth regions, and the mean local variance in the luminance component image includes the following steps: The image content complexity comprehensive characterization value is calculated based on the brightness component image. The comprehensive characterization value is jointly determined by the edge pixel ratio, the smooth region ratio, and the local variance mean. When the comprehensive characterization value is less than a preset lower limit, the current image is determined to be a low-complexity image; When the comprehensive characterization value is greater than the preset upper limit, the current image is determined to be a high-complexity image; When the comprehensive characterization value is between the preset lower limit and the preset upper limit, the current image is determined to be a medium complexity image.

[0010] In at least one embodiment of this application, the neighborhood size of the bilateral filter is set to 9, and the spatial weight parameter is fixed. The grayscale weight parameters of the bilateral filter are adjusted based on the content complexity parameters and the difference between the maximum and minimum brightness values ​​of the current frame's brightness component image, in order to improve denoising capability in low-complexity images and reduce edge blurring in high-complexity images.

[0011] In at least one embodiment of this application, the Canny edge detection includes the steps of: A low threshold is determined based on the content complexity parameter, and a high threshold is set to 2.5 times the low threshold. The low threshold increases as the content complexity parameter increases, in order to reduce noise edge response in high-complexity images.

[0012] In at least one embodiment of this application, the edge overlay includes the step of: The edge feature map is superimposed onto the brightness component image according to the enhancement weight to obtain the edge enhancement image; The enhancement weights are adjusted according to the content complexity parameter so that low-complexity images use smaller enhancement weights to suppress false edges, while high-complexity images use larger enhancement weights to enhance text, outlines, and structural lines.

[0013] In at least one embodiment of this application, the bicubic interpolation includes the steps of: Based on the pixel values ​​of the 16 neighboring pixels around the pixel to be interpolated in the edge enhancement image and the cubic interpolation weight function, the corresponding interpolated pixel value is calculated to generate a high-resolution enhanced image.

[0014] In at least one embodiment of this application, the Laplace sharpening includes the steps of: The high-resolution enhanced image is convolved using a convolution kernel with a center pixel coefficient greater than that of its neighboring pixels to enhance high-frequency information in the image. The sharpening intensity of the Laplacian sharpening is adjusted according to the content complexity parameter; Specifically, the bilateral filtering is performed before the Canny edge detection, the edge overlay is performed before the bicubic interpolation, and the Laplacian sharpening is performed after the bicubic interpolation.

[0015] A projector includes a processor and a memory, the memory storing a computer program that, when executed by the processor, implements the method for improving the image clarity of the projector as described above.

[0016] The above-mentioned method for improving the image quality of a projector involves the projector first obtaining a blank reference frame of the projection screen and constructing an illumination attenuation mask, and then performing illumination uniformity correction on the input projection image to improve the problem of local detail reduction caused by bright center and dark edge from the source. The corrected image is then converted to the YUV color space, and the luminance component is extracted as the main enhancement target to avoid distortion caused by over-processing the color information directly. Subsequently, the image content complexity parameter is determined based on the proportion of edge pixels, the proportion of smooth regions, and the mean of local variance in the brightness component image. This complexity parameter is then used to sequentially adjust bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening. This prioritizes noise suppression for images with more smooth backgrounds and prioritizes text and outline enhancement for images with more complex details. As a result, brightness balance, noise control, and sharpness enhancement are achieved simultaneously in different projection scenarios, reducing noise amplification, excessive edge enhancement, and artifact problems caused by simple sharpening. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the method for improving the image quality and clarity of a projector as described in this application. Figure 2 This is a schematic diagram of the media interaction process of the projector described in this application. Detailed Implementation

[0018] The embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0019] It should be noted that when a component is considered to be "connected" to another component, it can be directly connected to the other component or may also have an intervening component. When a component is considered to be "placed" on another component, it can be directly placed on the other component or may also have an intervening component. The terms "top," "bottom," "upper," "lower," "left," "right," "front," "back," and similar expressions used in this article are for illustrative purposes only.

[0020] Embodiments of this application provide a method for improving the image quality and clarity of a projector, the method comprising the following steps: S1. Obtain the input projection image and the blank reference frame of the projection screen, construct an illumination attenuation mask based on the blank reference frame, and use the illumination attenuation mask to perform illumination uniformity correction on the input projection image to obtain a corrected image. S2. Convert the corrected image into a YUV color space image to separate the luminance component image and the chrominance component image; S3. Determine the image content complexity parameters based on the proportion of edge pixels, the proportion of smooth regions, and the mean of local variance in the brightness component image. S4. Based on the complexity parameter, the brightness component image is sequentially subjected to bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening to obtain an output image with improved clarity.

[0021] The above-mentioned method for improving the image quality of a projector involves the projector first obtaining a blank reference frame of the projection screen and constructing an illumination attenuation mask, and then performing illumination uniformity correction on the input projection image to improve the problem of local detail reduction caused by bright center and dark edge from the source. The corrected image is then converted to the YUV color space, and the luminance component is extracted as the main enhancement target to avoid distortion caused by over-processing the color information directly. Subsequently, the image content complexity parameter is determined based on the proportion of edge pixels, the proportion of smooth regions, and the mean of local variance in the brightness component image. This complexity parameter is then used to sequentially adjust bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening. This prioritizes noise suppression for images with more smooth backgrounds and prioritizes text and outline enhancement for images with more complex details. As a result, brightness balance, noise control, and sharpness enhancement are achieved simultaneously in different projection scenarios, reducing noise amplification, excessive edge enhancement, and artifact problems caused by simple sharpening.

[0022] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0023] Please see Figures 1-2 The embodiments of this application provide a method for improving the image quality and clarity of a projector, and a projector in this embodiment, the method includes the following steps: S1. Obtain the input projection image and the blank reference frame of the projection screen, construct an illumination attenuation mask based on the blank reference frame, and use the illumination attenuation mask to perform illumination uniformity correction on the input projection image to obtain a corrected image.

[0024] Specifically, the input projection image to be processed and a blank reference frame of the projection screen are first acquired. The blank reference frame is preferably a reference image acquired when a uniform background is projected onto the screen under normal projection system conditions, which can reflect the illumination distribution characteristics of the current projection environment.

[0025] Further, the luminance distributions of the R, G, and B channels of the blank reference frame can be extracted respectively to generate corresponding R-channel attenuation masks, G-channel attenuation masks, and B-channel attenuation masks. Subsequently, based on the ratio of the original pixel values ​​of each color channel of the input projected image to the corresponding channel attenuation mask, normalization correction is performed on each color channel to obtain the corrected image. Preferably, the correction formula for each channel can be expressed as:

[0026]

[0027]

[0028] Where R(x,y), G(x,y), and B(x,y) represent the original pixel values ​​of the R, G, and B channels of the input image at coordinates (x,y), respectively. MR(x,y), MG(x,y), and MB(x,y) represent the illumination attenuation mask values ​​for the corresponding channels, respectively. R′(x,y), G′(x,y), and B′(x,y) represent the corrected pixel values ​​of each channel, respectively.

[0029] The concept of division correction—that is, using a pre-generated attenuation mask to correct the original image pixel by pixel to compensate for non-uniform illumination in projection displays or optical imaging systems—is common knowledge in the field of projection display uniformity correction. In multi-projector splicing and blending systems, generating a Luminance Attenuation Map (LAM) by measuring the projector's brightness attenuation characteristics and then performing correction is standard practice in this field. The reason for independent correction of each channel is that the optical attenuation characteristics of the R, G, and B channels in a real projection system vary due to differences in the projector's light source wavelength distribution, lens chromatic aberration, and screen reflection characteristics; therefore, it is usually necessary to generate an attenuation mask separately for each color channel.

[0030] The reason S1 is executed first is that the common problem of bright centers and dark edges in projected images is an inherent imbalance caused by the projection scene itself. If filtering, edge detection, and sharpening are performed directly on the uncorrected image, the already weak text and outline information in the dark edge areas will be further weakened. Therefore, this embodiment performs illumination uniformity correction first, so that subsequent enhancement is based on a more balanced brightness distribution, which is more conducive to the simultaneous improvement of the clarity of different areas.

[0031] S1 does not address image blurring itself, but rather the brightness imbalance unique to the projection scene. Only by first leveling the basic brightness can subsequent enhancement avoid unfairly amplifying dark areas at the edges, thus ensuring a balanced restoration of details in different areas. This is also one of the important preliminary steps in this embodiment compared to ordinary image enhancement schemes.

[0032] S2. Convert the corrected image into a YUV color space image to separate the luminance component image and the chrominance component image.

[0033] After obtaining the corrected image, the corrected image undergoes color space conversion, preferably to a YUV color space image, to separate the luminance component image and the chrominance component image. Preferably, the conversion formula can be expressed as: Y = 0.299R + 0.587G + 0.114B U= 0.147R 0.289G+0.436B V=0.615R 0.515G 0.100B Where R, G, and B are the input color components, Y is the luminance component, and U and V are the chrominance components.

[0034] This set of conversion coefficients comes from the BT.601 recommendation (Studio Digital Television Coding Parameters for Standard Definition Television) formulated by the International Telecommunication Union (ITU-R), and is the standard coefficient value for Y, Cb, and Cr (in this application, U and V have the same meaning as Cb and Cr) derived from the RGB space in standard definition television coding.

[0035] In this embodiment, the luminance component image is the primary target for subsequent sharpness enhancement, while the chrominance component image is preferably retained for final image reconstruction. This is because image sharpness, the legibility of text strokes, and edge transitions are mainly determined by luminance variations, while the chrominance component reflects more color information. Directly enhancing the chrominance component with the same intensity as the luminance component can easily amplify color noise and introduce color cast. Therefore, this embodiment focuses the enhancement on the luminance component image to minimize the risk of color distortion while improving sharpness.

[0036] S3. Based on the proportion of edge pixels, the proportion of smooth regions, and the mean of local variance in the brightness component image, determine the image content complexity parameters.

[0037] After obtaining the luminance component image, the image content complexity parameters are further determined based on the proportion of edge pixels, the proportion of smooth regions, and the mean of local variance in the luminance component image.

[0038] Among these, the edge pixel ratio characterizes the richness of high-frequency information such as contours, structural boundaries, and text strokes in the current image. The smooth region ratio characterizes the proportion of large, uniform background areas in the image. The mean local variance reflects the degree of local grayscale fluctuations and texture details in the image. By integrating these three types of information, a complexity parameter C can be obtained to characterize the complexity of the current image.

[0039] When the complexity parameter is small, the current image can be classified as a low-complexity image, such as an image with a smooth background mainly consisting of a sky, wall, or whiteboard. When the complexity parameter is large, the current image can be classified as a high-complexity image, such as an image containing text, charts, outlines, and complex textures. When the complexity parameter is between the two, it can be classified as a medium-complexity image.

[0040] The image content complexity parameter is a comprehensive representation value calculated based on the luminance component image. This comprehensive representation value is determined by the proportion of edge pixels, the proportion of smooth regions, and the mean local variance. Preferably, the image content complexity parameter C can be calculated using the following formula:

[0041] Among them, the edge pixel ratio reflects the richness of high-frequency information such as contour boundaries, text strokes, and structural lines in the current image; the smooth region ratio reflects the proportion of uniform background in the image; and the local variance mean reflects the amplitude of local grayscale changes and texture complexity. By comprehensively calculating the above three types of features, the content of the current image can be transformed from empirical judgment into quantifiable complexity parameters.

[0042] In one embodiment, the proportion of edge pixels can be obtained statistically based on edge detection results; the proportion of smooth regions can be obtained based on the proportion of regions where local grayscale changes are below a preset threshold; the mean of local variance can be obtained by calculating the local variance of the luminance component image using a preset window and taking the mean. When the complexity parameter C is less than a preset lower limit, the current image is determined to be a low-complexity image; when the complexity parameter C is greater than a preset upper limit, the current image is determined to be a high-complexity image; when the complexity parameter C is between the preset lower limit and the preset upper limit, the current image is determined to be a medium-complexity image.

[0043] After adopting the complexity parameter C mentioned above, the subsequent enhancement steps no longer use fixed parameters, but can be uniformly linked and controlled according to the complexity of the image content. This allows smooth background scenes to suppress noise first, while complex text and outline scenes to enhance details first, thus improving adaptability under different projection scenes.

[0044] S4. Based on the complexity parameters, the luminance component image is sequentially subjected to bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening to obtain an output image with improved clarity. After determining the complexity parameters, the multi-level image enhancement stage begins. The sub-steps in this step are not simply stacked, but have a clear input-processing-output logical relationship.

[0045] S41, Bilateral Filtering First, based on the complexity parameter, a bilateral filter is applied to the luminance component image to obtain a denoised luminance image. The advantage of bilateral filtering is that it considers not only the spatial distance between pixels but also the grayscale similarity between pixels, thus preserving edge structure as much as possible while suppressing noise. Its filtering formula can be expressed as:

[0046] Where I′(p) represents the filtered pixel value, I(p) represents the original pixel value, and q represents the neighboring pixels. Wp represents the normalization coefficient, Gs is the spatial Gaussian function, and Gr is the grayscale Gaussian function.

[0047] Gaussian weights in the spatial domain:

[0048] Gaussian weights in the grayscale region:

[0049] Bilateral filtering, proposed by Tomasi and Manduchi in 1998, is a classic algorithm in the field. Tomasi and Manduchi published a paper entitled "Bilateral Filtering for Gray and Color Images" at the 6th International Conference on Computer Vision (ICCV'98) in 1998 based on this work.

[0050] In at least one embodiment of this application, the neighborhood size of the bilateral filter is preferably set to 9, the spatial weight parameter sigmaSpace is fixed, and the grayscale weight parameter sigmaColor is adjusted according to the content complexity parameter C and the difference between the maximum and minimum brightness values ​​of the current frame's luminance component image. Preferably, sigmaColor can be determined by the following formula or a segmentation rule:

[0051] Where Ymax represents the maximum brightness value in the current frame's brightness component image, Ymin represents the minimum brightness value in the current frame's brightness component image, and α is a scaling factor, preferably ranging from 0.05 to 0.1.

[0052] The reason for adopting the aforementioned adaptive adjustment method of sigmaColor is that the grayscale weight parameter in bilateral filtering directly determines the smoothing degree for pixels of different grayscale levels. If sigmaColor is set too large, pixels with large grayscale differences will also be smoothed simultaneously, easily leading to blurring of text strokes and structural edges; if sigmaColor is set too small, the noise reduction capability is insufficient, and a large amount of background texture and compressed noise will still be retained. Therefore, a larger sigmaColor is preferred in low-complexity images to enhance the noise reduction capability in smoothing the background; a smaller sigmaColor is preferred in high-complexity images to reduce the blurring effect on true edges.

[0053] In this way, bilateral filtering not only achieves noise suppression, but also adaptively balances the relationship between denoising and edge preservation according to the complexity of the image content, providing a more suitable input image for subsequent edge detection.

[0054] First, curtain texture, wall grain, and compressed noise are removed from the luminance component image to obtain an input image more suitable for subsequent edge detection. If noise is not removed first, subsequent edge detection may misidentify noise as edges, thus affecting the overall enhancement effect.

[0055] S42, Canny edge detection Then, Canny edge detection is performed on the denoised brightness image to obtain an edge feature map. Preferably, the edge gradient magnitude can be expressed as:

[0056] Where Gx is the horizontal gradient and Gy is the vertical gradient; the edge direction can be represented as:

[0057] The Canny edge detection algorithm was published in 1986 by John F. Canny in the journal *IEEE Transactions on Pattern Analysis and Machine Intelligence*, titled "A Computational Approach to Edge Detection". This paper proposed a theoretically optimal edge detector model that maximizes the signal-to-noise ratio and optimizes localization accuracy, laying the theoretical foundation for modern edge detection and becoming one of the most cited classic papers in the field of computer vision. The unique feature of the Canny algorithm is its connection mechanism using high and low thresholds for hysteresis thresholding.

[0058] In at least one embodiment of this application, the low threshold of the Canny edge detection is adaptively determined based on the content complexity parameter C, and the high threshold is fixed at 2.5 times the low threshold, that is: high_threshold=2.5×low_threshold The low_threshold can be determined by the following formula or segmentation rule:

[0059] The reason for adopting the above threshold setting method is that in low-complexity images, the number of effective edges in the image is small. If the detection threshold is set too high, weak but meaningful edges are easily missed. In high-complexity images, if the threshold is still too low, a large number of false edges caused by texture, noise, or compression artifacts will be identified. Therefore, increasing the low threshold with the complexity parameter C can suppress noisy edge responses when the image complexity is high. At the same time, fixing the high threshold at 2.5 times the low threshold helps to improve the stability of edge screening while ensuring edge continuity.

[0060] By using the threshold linkage method described above, the edge detection step can be better adapted to different types of projected images, thereby improving the accuracy of text, contour and structural line extraction.

[0061] The edge information corresponding to text, contours and structural lines is extracted from the denoised brightness image to provide a more accurate edge feature map for subsequent edge enhancement.

[0062] S43, Edge Overlay After obtaining the edge feature map, the edge feature map is superimposed onto the brightness component image according to the enhancement weights to obtain the edge-enhanced image. Preferably, edge enhancement can be expressed as: I enh =I+αE Among them, Ienh Let I represent the image after edge enhancement, E represent the brightness component image, and α represent the enhancement weight.

[0063] It should be noted that this embodiment preferably employs a method of extracting edges from the denoised luminance image and then superimposing the edge feature map onto the original luminance component image. This is because the denoised luminance image is more suitable for extracting true edges, while the original luminance component image retains a more complete grayscale level and detail base. In this way, on the one hand, the interference of noise on edge extraction can be reduced, and on the other hand, the true edges can be enhanced while preserving the original luminance level, thus achieving a balance between detail preservation and improved edge sharpness.

[0064] In some implementations, the enhancement weights are adjusted based on the content complexity parameter so that low-complexity images use smaller enhancement weights to suppress false edges, while high-complexity images use larger enhancement weights to enhance text, outlines, and structural lines.

[0065] In at least one embodiment of this application, the edge enhancement weight α is adaptively adjusted according to the content complexity parameter C. Preferably, the enhancement weight α can be determined by the following formula or a segmentation rule:

[0066] The aforementioned adaptive enhancement weight setting method is employed because the edge feature map overlay process essentially involves reintegrating high-frequency contour information into the luminance component image. If the enhancement weight is set too high, false edges, white edges, or local overshoot may easily appear in smooth background areas; if the enhancement weight is set too low, it will be difficult to adequately highlight text, contours, and structural lines in highly complex images. Therefore, a smaller enhancement weight is preferred in low-complexity images to suppress false enhancements in smooth areas; a larger enhancement weight is preferred in highly complex images to enhance the representation of true edges and details.

[0067] It should be noted that the edge feature map in this application is preferably derived from the denoised luminance image, while the superimposed object is preferably the original luminance component image. This makes the denoised luminance image more suitable for accurately extracting real edges, while the original luminance component image retains more complete grayscale levels and luminance details. By lifting edges from the denoised image and padding edges on the original luminance image, noise interference can be reduced while preserving the original luminance levels, thus achieving a more natural edge enhancement effect.

[0068] S44, Bicubic Interpolation Subsequently, bicubic interpolation is performed on the edge enhancement image to obtain a high-resolution enhanced image. Preferably, bicubic interpolation calculates the interpolated pixel value using the 16 neighboring pixels surrounding the pixel to be interpolated, and its higher-order expression can be represented as:

[0069] Where I(x+i,y+j) represents the neighboring pixel value, and w(i) represents the cubic interpolation weight function.

[0070] Bicubic interpolation (also known as Catmull-Rom spline interpolation) was proposed by Catmull and Rom in 1974. The algorithm was first included in the collection of papers *Computer Aided Geometric Design*, titled "A class of local interpolating splines". The core weight function of bicubic interpolation is derived from the theory of Catmull-Rom spline interpolation and is a standard algorithm in computer graphics for image upscaling and texture mapping.

[0071] The resolution is improved on the basis of the already enhanced edges, thereby compensating for the resolution loss caused by projection imaging and reducing the problems of false edges, white edges and ringing that may be caused by enlarging and then enhancing.

[0072] In at least one embodiment of this application, the bicubic interpolation is a resolution enhancement process performed on an edge-enhanced image, rather than a direct enlargement of the original brightness image or a denoised brightness image.

[0073] If interpolation and magnification are performed first, followed by edge enhancement, the pseudo-edges generated during interpolation will be further magnified by subsequent enhancement, easily causing white edges and ringing. However, if edge enhancement is performed first, followed by bicubic interpolation, the strokes, outlines, and structural lines that truly need strengthening can be extended more smoothly during the magnification process, thereby improving the naturalness and continuity of the high-resolution output image.

[0074] The high-resolution enhanced image obtained through this step not only has a higher display resolution but also retains the contour enhancement effect brought about by the edge enhancement in the previous step, providing a more suitable input basis for the final sharpening process. S45, Laplace Sharpening Finally, the high-resolution enhanced image is subjected to Laplacian sharpening to obtain an output image with improved clarity. Preferably, a convolution kernel with a center pixel coefficient greater than that of adjacent pixels can be used, for example:

[0075] The corresponding sharpening expression can also be written as:

[0076] Among them, I sharpG(I) represents the sharpened image, and k represents the sharpening coefficient.

[0077] Laplacian sharpening is a standard method in digital image processing for enhancing image edges and high-frequency information. The Laplacian operator, as a second-order differential operator, is detailed in Section 3.7.2, "Sharpening Spatial Filters," of Gonzalez and Woods' *Digital Image Processing* (2nd edition, starting page 128). The mathematical essence of edge enhancement is to proportionally superimpose the original image with the result of a high-pass filter, thereby significantly improving sharpness by enhancing high-frequency components (edges and details).

[0078] In one implementation, the sharpening factor is adjusted according to the content complexity parameter to avoid over-sharpening noise in low-complexity images and to highlight text and outline details in high-complexity images.

[0079] The reason sharpening is placed as the last step is that it enhances the high-frequency information of the image. If sharpening is performed too early, it will amplify any noise that has not yet been cleared, and this noise will be further amplified during subsequent interpolation and stacking processes. Therefore, in this embodiment, sharpening is preferably used as the final step in the entire enhancement chain.

[0080] In at least one embodiment of this application, the sharpening coefficient k in the Laplacian sharpening is adaptively adjusted according to the content complexity parameter C. Preferably, the sharpening coefficient k can be determined by the following formula or a segmentation rule:

[0081] Laplacian sharpening essentially enhances the high-frequency components of an image. If a large sharpening factor is used in low-complexity images, residual noise in smooth background areas is easily amplified; conversely, if the sharpening factor is too small in high-complexity images, the enhanced text, contours, and structural lines are difficult to highlight further. Therefore, a smaller sharpening factor is preferred in low-complexity images to suppress noise amplification in smooth areas, while a larger sharpening factor is preferred in high-complexity images to further enhance text edges and details.

[0082] Furthermore, this application preferably sets Laplacian sharpening to be performed after bicubic interpolation. The advantage of this is that when sharpening is applied to a high-resolution enhanced image, it allows for more focused finishing enhancement of the recovered contours and details, avoiding premature amplification of noise in the front-end processing. Thus, the final output image can simultaneously achieve high resolution, sharper edge transitions, and a lower risk of artifacts.

[0083] It should be noted that the execution order of each step in this application has a clear technical dependency and is not arbitrarily arranged. Preferably, the overall execution order is: illumination uniformity correction, color space conversion, complexity parameter determination, bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening.

[0084] The process involves several steps: first, performing illumination uniformity correction to eliminate interference from the center being bright and the edges being dark in the projected scene; second, performing color space conversion and extracting the luminance component to focus the sharpness enhancement on luminance information and reduce color distortion; third, performing bilateral filtering before edge detection to reduce the risk of noise being misidentified as edges; fourth, performing edge overlay before bicubic interpolation to avoid white edges and ringing caused by enlarging the image before patching the edges; and finally, performing Laplacian sharpening to finish the enhancement on a high-resolution basis and prevent front-end noise from being amplified prematurely.

[0085] By arranging the above sequence, this application can better balance brightness balance, noise suppression and detail enhancement in projection scenarios, thereby reducing the noise amplification, excessive edge enhancement and artifact problems common in simple sharpening schemes.

[0086] S46, Final Output Image In this embodiment, the output image with enhanced clarity can be formed by recombining the enhanced luminance component image and chrominance component image; when an RGB format image needs to be output, the combined YUV image can be further converted into an RGB image for output. This can highlight text, outlines, and structural details while maintaining the stability of color information.

[0087] In at least one embodiment of this application, the chroma component image preferably remains unchanged during the enhancement process of the luminance component image. After the luminance component image has undergone bilateral filtering, edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening, the enhanced luminance component image is combined with the corresponding chroma component image to form an output YUV image. When an RGB format image needs to be output, the combined YUV image can be further subjected to inverse color space conversion to form an RGB output image.

[0088] This application focuses its enhancement of sharpness primarily on the luminance component, without applying the same level of enhancement to the chrominance component. This approach aims to improve image sharpness while minimizing issues such as color noise amplification, color distortion, and color fringing. By recombinating the enhanced luminance and chrominance components at the final output stage, both improved sharpness and color stability can be achieved.

[0089] Therefore, the above-mentioned method for improving the image quality of a projector and the projector first obtain a blank reference frame of the projection screen and construct an illumination attenuation mask, and then perform illumination uniformity correction on the input projection image to improve the problem of local detail reduction caused by bright center and dark edge from the source. The corrected image is then converted to the YUV color space, and the luminance component is extracted as the main enhancement target to avoid distortion caused by over-processing the color information directly. Subsequently, the image content complexity parameter is determined based on the proportion of edge pixels, the proportion of smooth regions, and the mean of local variance in the brightness component image. This complexity parameter is then used to sequentially adjust bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening. This prioritizes noise suppression for images with more smooth backgrounds and prioritizes text and outline enhancement for images with more complex details. As a result, brightness balance, noise control, and sharpness enhancement are achieved simultaneously in different projection scenarios, reducing noise amplification, excessive edge enhancement, and artifact problems caused by simple sharpening.

[0090] The above description is merely an embodiment of this application. It should be noted that those skilled in the art can make improvements without departing from the inventive concept of this application, but these improvements all fall within the protection scope of this application.

Claims

1. A method for improving the image clarity of a projector, characterized in that, Includes the following steps: Obtain the input projection image and the blank reference frame of the projection screen. Construct an illumination attenuation mask based on the blank reference frame. Use the illumination attenuation mask to perform illumination uniformity correction on the input projection image to obtain the corrected image. The corrected image is converted into a YUV color space image to separate the luminance component image and the chrominance component image; Based on the proportion of edge pixels, the proportion of smooth regions, and the mean local variance in the brightness component image, the image content complexity parameter is determined. Based on the complexity parameters, the luminance component image is sequentially subjected to bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening to obtain an output image with improved clarity.

2. The method for improving the image clarity of a projector according to claim 1, characterized in that, The steps of sequentially performing bilateral filtering, Canny edge detection, edge overlay, bicubic interpolation, and Laplacian sharpening on the luminance component based on the complexity parameter to obtain an output image with improved clarity include: Based on the content complexity parameter, the luminance component image is subjected to bilateral filtering to obtain a denoised luminance image; Canny edge detection is performed on the denoised brightness image to obtain an edge feature map; The edge feature map is superimposed on the deluminance component image to obtain the edge-enhanced image; Bicubic interpolation is performed on the edge enhancement image to obtain a high-resolution enhancement image; The high-resolution enhanced image is then subjected to Laplacian sharpening to obtain an output image with improved clarity.

3. The method for improving the image clarity of a projector according to claim 1, characterized in that, The step of constructing an illumination attenuation mask based on the blank reference frame and using the illumination attenuation mask to perform illumination uniformity correction on the input projected image includes the following steps: The brightness distributions of the R, G, and B channels of the blank reference frame are extracted respectively to generate corresponding R channel attenuation masks, G channel attenuation masks, and B channel attenuation masks; The input projected image is normalized and corrected by the ratio of the original pixel value of each channel to the corresponding channel attenuation mask.

4. The method for improving the image clarity of a projector according to claim 1, characterized in that, The step of determining the image content complexity parameter based on the proportion of edge pixels, the proportion of smooth regions, and the mean local variance in the brightness component image includes the following steps: The image content complexity comprehensive characterization value is calculated based on the brightness component image. The comprehensive characterization value is jointly determined by the edge pixel ratio, the smooth region ratio, and the local variance mean. When the comprehensive characterization value is less than a preset lower limit, the current image is determined to be a low-complexity image; When the comprehensive characterization value is greater than the preset upper limit, the current image is determined to be a high-complexity image; When the comprehensive characterization value is between the preset lower limit and the preset upper limit, the current image is determined to be a medium complexity image.

5. The method for improving the image clarity of a projector according to claim 4, characterized in that, The neighborhood size of the bilateral filter is set to 9, and the spatial weight parameter is fixed. The grayscale weight parameters of the bilateral filter are adjusted based on the content complexity parameters and the difference between the maximum and minimum brightness values ​​of the current frame's brightness component image, in order to improve denoising capability in low-complexity images and reduce edge blurring in high-complexity images.

6. The method for improving the image clarity of a projector according to claim 4, characterized in that, The Canny edge detection includes the following steps: A low threshold is determined based on the content complexity parameter, and a high threshold is set to 2.5 times the low threshold. The low threshold increases as the content complexity parameter increases, in order to reduce noise edge response in high-complexity images.

7. The method for improving the image clarity of a projector according to claim 4, characterized in that, The edge overlay includes the following steps: The edge feature map is superimposed onto the brightness component image according to the enhancement weight to obtain the edge enhancement image; The enhancement weights are adjusted according to the content complexity parameter so that the enhancement weights for low-complexity images suppress false edges, while the enhancement weights for high-complexity images enhance text, outlines, and structural lines.

8. The method for improving the image clarity of a projector according to claim 7, characterized in that, The bicubic interpolation includes the following steps: Based on the pixel values ​​of the 16 neighboring pixels around the pixel to be interpolated in the edge enhancement image and the cubic interpolation weight function, the corresponding interpolated pixel value is calculated to generate a high-resolution enhanced image.

9. The method for improving the image clarity of a projector according to claim 8, characterized in that, The Laplace sharpening process includes the following steps: The high-resolution enhanced image is convolved using a convolution kernel with a center pixel coefficient greater than that of its neighboring pixels to enhance high-frequency information in the image. The sharpening intensity of the Laplacian sharpening is adjusted according to the content complexity parameter; Specifically, the bilateral filtering is performed before the Canny edge detection, the edge overlay is performed before the bicubic interpolation, and the Laplacian sharpening is performed after the bicubic interpolation.

10. A projector, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, which, when executed by the processor, implements the method for improving the image clarity of a projector as described in any one of claims 1-9.

Images