A Fast Image Super-Resolution Reconstruction Method Based on Four-Pixel Root Mean Square Function
By employing a fast image super-resolution reconstruction method based on the four-pixel root mean square function, and utilizing linear transformation and sliding window sweeping techniques combined with statistical principles, this method solves the problems of high computational cost and detail smoothing in existing methods, achieving fast and efficient high-resolution image reconstruction.
Patent Information
- Application Number
- CN202310198232.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-03-03
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2043-03-03
AI Technical Summary
Existing image super-resolution reconstruction methods based on the four-pixel root mean square function are computationally intensive and converge slowly. High-resolution image details are easily smoothed out, making it difficult to obtain high-detail images with fast processing and low computational cost.
A fast image super-resolution reconstruction method based on the four-pixel root mean square function is adopted. The four-pixel bounding box is constructed through linear transformation and sliding window. The root mean square function is used for sweeping and gray value inversion transformation. Combined with statistical principles, it is transformed into a well-posed reconstruction problem, which preserves image details and reduces the amount of computation.
It achieves high-fidelity image reconstruction with fast processing and low computational cost, and the image remains smooth and undistorted after magnification, with clear details.
Smart Images

Figure FT_1 
Figure FT_2 
Figure SMS_1
Abstract
Description
Technical Field
[0001] This invention relates to the field of image processing, particularly to the field of image super-resolution reconstruction, and specifically to a fast image super-resolution reconstruction method based on the four-pixel root mean square function. Background Technology
[0002] Resolution is one of the important indicators for evaluating image quality. Higher resolution images mean that they can provide richer detail information and have better visual effects and image quality. However, in reality, due to the limitations of imaging system hardware, as well as the influence of factors such as noise and focus deviation, the images we obtain are often low in resolution, cannot meet the needs of practical applications, and have noise and blurring.
[0003] Image super-resolution reconstruction utilizes existing equipment and signal processing techniques to overcome the inherent resolution limitations of imaging systems through software means. It reconstructs high-resolution images from low-resolution images, thereby improving image quality and visual effects. Specifically, it can mitigate the spatial resolution degradation caused by image downscaling and discretization, tap the potential of existing image data (such as multi-angle, multi-temporal, multi-platform remote sensing images, and image sequences), break through the spatial resolution limits of image acquisition methods, and achieve significant results. It is convenient to apply and inexpensive.
[0004] Currently, in the research of super-resolution image reconstruction technology, regularization-based reconstruction methods and statistical reconstruction methods are the most commonly used and effective mainstream methods.
[0005] Furthermore, current market-based fast image super-resolution reconstruction methods based on the four-pixel root mean square function combine prior knowledge of the image and fuzziness to address the problem. They solve for the minimum functional based on the forward model of the degraded image sequence to obtain a high-resolution image. However, these methods are computationally intensive, converge slowly, and the details of the high-resolution image are easily smoothed out. Summary of the Invention
[0006] The present invention aims to provide a fast image super-resolution reconstruction method based on the four-pixel root mean square function. This method can overcome the above-mentioned shortcomings of the prior art and can obtain an infinitely magnified image with high detail fidelity while being fast and computationally inefficient.
[0007] The technical solution of the present invention is as follows: The image super-resolution fast reconstruction method based on the four-pixel root mean square function includes the following steps: A. Perform a linear transformation on the input image, subtracting the maximum gray value of all pixels in the input image from the set value, thereby swapping the brighter and darker areas in the input image and completing the inversion transformation of the input image. B. Create a 2×2 sliding window as a four-pixel bounding box, and construct a four-pixel root mean square function, as shown in the following formula: (1) Where n is the vertical magnification factor of the pixel, t is the horizontal magnification factor of the pixel, x is the square root exponent, A is the gray level of the top left pixel of the four-pixel box, B is the gray level of the top right pixel of the four-pixel box, C is the gray level of the bottom left pixel of the four-pixel box, and D is the gray level of the bottom right pixel of the four-pixel box. C. Sweep the input image using a four-pixel bounding box to obtain the processed image; D. Invert the grayscale values of the image to be processed. Subtract the maximum grayscale value from all pixels in the input image to swap the brighter and darker areas, and obtain the final output result.
[0008] In the formula (1), x takes the value n*t / y, and y is a value between 3 and 8.
[0009] In step C, the sweeping process is as follows: starting from the top left or top right corner, shift one pixel at a time. After reaching the other end, perform a line break, and then shift one pixel in the opposite direction. This process is repeated until the entire image is processed to obtain the processed image.
[0010] The method of this invention utilizes statistical principles to treat the super-resolution reconstruction problem as a statistical estimation problem. By applying the root mean square function in probability statistics, it maximizes the posterior probability of the occurrence of a high-resolution image given a known low-resolution image sequence, thus transforming the ill-posed reconstruction problem into a well-posed reconstruction problem. This achieves high-resolution image reconstruction with both data fidelity and edge preservation.
[0011] The method of this invention is a fast image super-resolution reconstruction method based on the four-pixel root mean square function. It is a smaller metric than pixels and uses the changes in pixel grayscale values to obtain the pixel change trend.
[0012] This invention proposes a new four-pixel root mean square function for grayscale value calculation. This formula can minimize the amount of calculation at the same magnification while ensuring smoothness and no distortion after magnification, and achieve the best effect at the same amount of calculation, thereby obtaining a vectorized image that retains details and is smooth. Attached Figure Description
[0013] Figure 1 The original image as input; Figure 2 This is an enlarged version of the original image. Detailed Implementation
[0014] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0015] Example 1 The image super-resolution fast reconstruction method based on the four-pixel root mean square function includes the following steps: The image super-resolution fast reconstruction method based on the four-pixel root mean square function includes the following steps: A. Perform a linear transformation on the input image, subtracting all pixels in the input image from the set maximum gray value, thereby swapping the brighter and darker areas in the input image and completing the inversion transformation of the input image; in this embodiment, the maximum gray value is set to 256. B. Create a 2×2 sliding window as a four-pixel bounding box, and construct a four-pixel root mean square function, as shown in the following formula: (1) Where n is the vertical magnification factor of the pixel, which is 100 in this embodiment; t is the horizontal magnification factor of the pixel, which is 100 in this embodiment; x is the square root exponent, which is n*t / 6 in this embodiment; A is the gray level of the top left pixel of the four-pixel box; B is the gray level of the top right pixel of the four-pixel box; C is the gray level of the bottom left pixel of the four-pixel box; and D is the gray level of the bottom right pixel of the four-pixel box. C. Use a four-pixel box to sweep the input image. The sweeping process is as follows: start from the top left or top right corner, shift one pixel at a time, and after reaching the other end, perform a line break. Then shift one pixel in the opposite direction, and so on, to process the entire image and obtain the processed image. D. Invert the grayscale values of the image to be processed. Subtract the maximum grayscale value from all pixels in the input image to swap the brighter and darker areas, and obtain the final output result.
[0016] Example 2 The plastic part image was magnified using the method in Example 1. See the attached image for details. Figure 1 and Figure 2 , Figure 1 Original image of the part. Figure 2 This is an enlarged image.
[0017] from Figure 1 and Figure 2 The comparison shows that, Figure 1 When magnified, the image becomes clearer, smoother, and without distortion.
Claims
1. A fast image super-resolution reconstruction method based on a four-pixel root mean square function, characterized in that, Includes the following steps: A. Perform a linear transformation on the input image, subtracting the maximum gray value of all pixels in the input image from the set value, thereby swapping the brighter and darker areas in the input image and completing the inversion transformation of the input image. B. Create a 2×2 sliding window as a four-pixel bounding box, and construct a four-pixel root mean square function, as shown in the following formula: (1) Where n is the vertical magnification factor of the pixel, t is the horizontal magnification factor of the pixel, x is the square root exponent, A is the gray level of the top left pixel of the four-pixel box, B is the gray level of the top right pixel of the four-pixel box, C is the gray level of the bottom left pixel of the four-pixel box, and D is the gray level of the bottom right pixel of the four-pixel box. C. Sweep the input image using a four-pixel bounding box to obtain the processed image; D. Invert the grayscale values of the image to be processed. Subtract the maximum grayscale value from all pixels in the input image to swap the brighter and darker areas, and obtain the final output result.
2. The image super-resolution fast reconstruction method based on the four-pixel root mean square function as described in claim 1, characterized in that: In the formula (1), x takes the value of n*t / y, and y is a value between 3 and 8.
3. The image super-resolution fast reconstruction method based on the four-pixel root mean square function as described in claim 1, characterized in that: In step C, the sweeping process is as follows: starting from the top left or top right corner, shift one pixel at a time. After reaching the other end, perform a line break, and then shift one pixel in the opposite direction. This process is repeated until the entire image is processed to obtain the processed image.
Citation Information
Patent Citations
Method for reestablishment of single frame image quick super-resolution based on nucleus regression
CN101226634A
Plant protection unmanned aerial vehicle ground monitoring system and monitoring method thereof
CN111860375A
