Spatial target image enhancement method under low illumination intensity, program product and medium

Abstract

The invention relates to an image enhancement method, in particular to a space target image enhancement method under low illumination intensity, a program product and a medium, and aims to overcome the defect that in the prior art, a space target is difficult to identify under the low illumination condition, and the space target image enhancement method under the low illumination intensity comprises the following steps: reading an original image; converting a color space, generating a binary image, performing contour extraction on the binary image, and selecting an outer contour communication region with the maximum area as an initial target region; performing adaptive gamma correction on the initial target area to obtain a corrected color component image; and generating a synthetic mask, performing image fusion by using the synthetic mask, the original image and the corrected color component image to realize secondary enhancement effect optimization, and merging channels to obtain an enhanced space target image. Meanwhile, the invention further provides a computer program product and a readable storage medium capable of realizing the method.

Description

Technical Field

[0001] The present invention relates to an image enhancement method, and specifically to an image enhancement method, program product and medium for spatial target images under low light intensity. Background Art

[0002] Humans use camera devices to photograph spatial objects, so as to more intuitively understand the spatial appearance in the form of images. However, during the image acquisition process, the spatial lighting environment is complex, including backlighting and side lighting. In addition, there may be insufficient lighting under meteorite occlusion. These factors lead to the degradation of the acquired images and the unclear outlines of spatial targets. Therefore, it is necessary to solve the problem of enhancing the outlines of image targets to support subsequent tasks under low light intensity, such as target tracking and non-cooperative target attitude estimation in space applications.

[0003] To improve the imaging quality of the space environment under low illuminance, a practical low-light high-quality imaging solution is essential. Traditional image enhancement algorithms, such as histogram equalization, image sharpening, and gamma correction, usually adopt global brightness and contrast enhancement strategies, but do not fully consider the non-uniformity of light distribution, resulting in poor restoration effects in target areas and low universality, making it difficult to identify spatial targets under low light conditions. In recent years, image enhancement methods based on deep learning have made significant progress, such as super-resolution convolutional neural network (SRCNN), feed-forward denoising neural network (DnCNN), and unsupervised generative adversarial network (EnlightenGAN). However, these methods still face certain challenges in practical applications, and their maturity and practicality need to be further improved. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiency in the prior art that it is difficult to identify spatial targets under low light conditions, and to provide an image enhancement method, program product and medium for spatial target images under low light intensity, so as to improve the accuracy of target contour recognition under low light intensity and when the spatial target is similar to the background color.

[0005] To achieve the above purpose, the technical solution provided by the present invention is as follows: An image enhancement method for spatial target images under low light intensity, characterized in that it includes the following steps: Step 1, read in the original image; Step 2, convert the original image into the YCbCr color space, perform threshold segmentation to generate a binary image; Step 3, perform edge detection on the binary image to obtain the result image of edge detection, merge it with the binary image to obtain a binary result image, and filter out isolated noise points in the binary result image; Step 4: Perform closing and opening morphological operations on the binary result image after noise filtering in sequence; Step 5: Extract all the contours in the binary result image processed in Step 4, calculate the area of the connected region enclosed by each contour, and select the connected region with the largest area as the initial target region; Step 6: Fill the small holes in the initial target region to obtain the binary image of the complete target region , where is the pixel position index; Step 7: Perform adaptive gamma correction on the R, G, and B color component images of the original image respectively to obtain the corrected color component images , where represents the color component, ; Step 8: Obtain the synthesis mask according to the binary image of the complete target region ; Step 9: Use the synthesis mask , the R, G, and B color component images of the original image and the corrected color component images to perform image fusion to obtain the fused image , merge the color channels, and complete the enhancement of the spatial target image.

[0006] Furthermore, Step 7 is specifically as follows: Decompose the original image into R, G, and B three color components to obtain the corresponding color component images , and adaptively derive the parameter according to the average gray value of each color component, as shown below:

[0007] where are the height and width of the original image respectively; Then, perform correction on the color component image through the adaptive gamma correction function to obtain the corrected color component image , as shown below:

[0008] .

[0009] Furthermore, Step 8 is specifically as follows: Step 8.1: Detect the binary image of the complete target region The boundary is processed by Gaussian blurring to obtain a blurred boundary image of the complete target area. ; Step 8.2. Obtain the composite mask through the following formula : .

[0010] Furthermore, in step 9, use , and to perform image fusion to obtain a fused image , specifically: .

[0011] Furthermore, in step 2, the threshold segmentation is specifically: take the luminance component map, and perform threshold segmentation according to the mean value of the luminance component map; In step 3, the edge detection uses Canny edge detection; use median filtering to filter out isolated noise points in the binary result map; Step 4 is specifically: create an elliptical kernel with a radius of 19 - 22 pixels as a structural element; then use the elliptical kernel to perform closing and opening morphological operations on the binary result map with noise points filtered out in sequence.

[0012] Furthermore, in step 3, the merging method is OR; In step 4, create an elliptical kernel with a radius of 20 pixels as a structural element.

[0013] Meanwhile, the present invention also provides a computer program product, including a computer program, characterized in that: when the program is executed by a processor, it implements the steps of the above - mentioned spatial target image enhancement method under low light intensity.

[0014] The present invention also provides a computer - readable storage medium, on which a computer program is stored, characterized in that: when the program is executed by a processor, it implements the steps of the above - mentioned spatial target image enhancement method under low light intensity.

[0015] Advantages of the present invention: 1. The present invention comprehensively applies technologies such as threshold segmentation, contour extraction, adaptive gamma correction, and image fusion, effectively retains texture details while enhancing the spatial target, restores the color to be relatively natural, has a simple process, and has high computational efficiency and significant enhancement effect.

[0016] 2. The present invention uses an adaptive gamma correction function to enhance the brightness of the target area, and combines a synthetic mask for image fusion to further optimize the correction result, thereby effectively restoring more detailed texture and contour information, improving the identifiability of the target area, and further enhancing the accuracy of spatial target detection, pose estimation, etc. in subsequent applications.

[0017] 3. The present invention performs image fusion on the synthetic mask, the R, G, and B color component images of the original image, and the corrected color component images, which can effectively smooth the color transition at the edge of the target area, and finally obtain a more natural low-illumination spatial target image enhancement effect.

[0018] 4. The present invention also provides a computer program product and a computer-readable storage medium capable of executing the above method steps, which can popularize and apply the method of the present invention to achieve spatial target image enhancement on corresponding hardware devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic flowchart of an embodiment of the method for enhancing a spatial target image under low light intensity according to the present invention; Figure 2 It is a spatial target contour diagram obtained in an embodiment of the method for enhancing a spatial target image under low light intensity according to the present invention; Figure 3 It is a schematic diagram of the synthetic mask obtained in step ten of an embodiment of the method for enhancing a spatial target image under low light intensity according to the present invention; Figure 4 It is a comparison diagram before and after enhancing a spatial target image in an embodiment of the method for enhancing a spatial target image under low light intensity according to the present invention, where (a) is before enhancement and (b) is after enhancement. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] The method for enhancing a spatial target image under low light intensity according to the present invention realizes the full-automatic contour extraction of the spatial target and the adaptive enhancement of the target area. First, the image is converted from the RGB color space to the YCbCr color space, the Y luminance component map is extracted, and then a threshold is set according to the mean value of the luminance component to generate a binary map and extract the target contour. The specific process is as Figure 1 shown, including the following steps: Step 1: Read in the original image.

[0021] Step 2: Convert the read original image from an RGB image to the YCbCr color space, and take the luminance component map.

[0022] Step 3: Calculate the mean value of the luminance component map, use it as the threshold, perform threshold segmentation, and generate a binary map.

[0023] Step 4: Apply Canny edge detection to process the binary image, obtaining the spatial target contour in a low light intensity environment, as Figure 2 shown, thereby obtaining the result image of edge detection.

[0024] In this embodiment, the effect of using Canny edge detection is good, with a high signal-to-noise ratio, high positioning accuracy, and strong single-edge response ability. In other embodiments, other edge detection methods such as Sobel edge detection, Roberts edge detection, Marr-Hildreth edge detection, Laplacian of Gaussian edge detection, and Transformer-based edge detection can also be used.

[0025] Step 5: Merge the result image of edge detection with the binary image obtained in Step 3 to obtain the final binary result image. Subsequently, apply median filtering to it to filter out isolated noise points.

[0026] The method of merging the result image of edge detection with the binary image obtained in Step 3 is OR operation. Specifically: the binary image consists of 0 and 1, and the edge detection result is also a binary image. The OR operation means performing a bitwise OR operation on the pixel values at the corresponding positions. If at least one of the two is 1, the result is 1; if both are 0, the result is 0, that is, 0 OR 0 is 0, 0 OR 1 is 1, and 1 OR 1 is 1.

[0027] Median filtering is a non-linear smoothing technique that sets the gray value of the central pixel point to the median of the gray values of all pixel points within the neighborhood window of this point. It has a good filtering effect on "salt-and-pepper noise" and isolated noise points. In this embodiment, the size of the neighborhood window is 20×20 pixel points, and the pixel values of the filled pixel points at the image boundary are defaulted to 0. Specifically, median filtering selects the central pixel point of the digital image and the pixel points included in its neighborhood window, sorts these pixel points according to the pixel value size, and then takes the pixel value at the middle position as the pixel value of the current central pixel point, thereby eliminating isolated noise points.

[0028] Step 6: Perform closing and opening morphological operations on the binary result image after filtering out noise points in sequence. Specifically: create a structuring element. In this embodiment, the structuring element is an elliptical kernel with a radius of 20 pixels; then use the elliptical kernel to perform closing and opening morphological operations on the binary result image after filtering out noise points in sequence. In other embodiments of the present invention, other elliptical kernels with radii in the range of 19 - 22 pixels can also be selected.

[0029] The closing operation first performs a dilation operation and then an erosion operation on the binary result image, mainly used to fill small holes in the region and connect the disconnected parts between the targets; the opening operation is the opposite of the closing operation, first performing an erosion operation and then a dilation operation, mainly used to remove small objects or noise in the image, smooth the boundaries of the targets, and remove small protrusions at the edges. This operation sequence can effectively ensure the integrity of the target area.

[0030] Step Seven: Extract all the contours in the binary result image processed in Step Six, calculate the area of the connected region enclosed by each contour in pixels, and select the connected region with the largest area as the initial target area.

[0031] Step Eight: Fill the small holes in the initial target area to achieve complete recognition of the target area and obtain the binary image of the complete target area. .

[0032] Step Nine: Through the adaptive gamma correction function, correct the R, G, and B color component images of the original image to obtain the corrected color component images , where respectively represent the R, G, and B color channels. The adaptive gamma correction function of the present invention is defined as , where the set parameter is adaptively derived from the average gray value of each color component. Associate the R, G, and B color component images with the adaptive gamma correction function to improve the correction effect. Specifically as follows:

[0033]

[0034]

[0035] where are the height and width of the original image respectively.

[0036] Step Ten: Due to the limited target segmentation accuracy, after adaptive gamma correction, the edge processing effect of the target area is not ideal. Therefore, the present invention proposes a method for generating a synthetic mask , specifically: perform Canny edge detection on the binary image of the complete target area obtained in Step Eight , use the Gaussian function for filtering processing, that is, perform Gaussian blur processing to smooth the edge transition, and obtain the blurred boundary image of the complete target area , and then use to obtain the synthetic mask , as shown in Figure 3 , and the expression is as follows:

[0037] Among them, is the pixel position index.

[0038] Step Eleven: Optimize the enhancement effect of the spatial target area. Use the synthetic mask, the R, G, and B color component images of the original image, and the corrected color component images for image fusion to obtain a fused image , expressed as:

[0039] Subsequently, merge the color channels to complete the enhancement of the spatial target image and obtain the enhanced spatial target image. As shown in (b) of Figure 4 , compared with (a) of Figure 4 , it can be seen that the enhancement effect of the present invention is significant.

[0040] The spatial target image enhancement method under low light intensity of the present invention can also form a computer program product. The program product includes a computer program, and when the program is executed by a processor, it implements the steps of the spatial target image enhancement method under low light intensity. In addition, the spatial target image enhancement method of the present invention can be applied in a computer-readable storage medium. The computer-readable storage medium stores a computer program, and the above spatial target image enhancement method can be stored in the computer-readable storage medium as a computer program. When the computer program is executed by a processor, it implements the above steps of the spatial target image enhancement method under low light intensity.

Claims

1. A method for enhancing space target images under low light intensity, characterized in that: The following steps are involved: Step 1, read the original image; Step 2, convert the original image into YCbCr color space, perform threshold segmentation, and generate a binary image; Step 3, performing edge detection on the binary image to obtain an edge detection result image, merging the edge detection result image with the binary image to obtain a binary result image, and filtering out isolated noise points in the binary result image; Step 4, performing closing operation and opening operation morphological operations on the binary result image after filtering out noise points; Step 5, extract all contours in the binary result image processed by step 4, calculate the area of ​​the connected region surrounded by each contour, and select the connected region with the largest area as the initial target region; Step 6: Fill the small holes in the initial target area to obtain a complete binary image of the target area. ,in, is the pixel position index; Step 7: The three color component images of R, G, and B of the original image Perform adaptive gamma correction respectively to obtain the corrected color component images ,in, Represents the color component, ; Step 8: Based on the complete target area binary image Get the synthetic mask ; Step 9: Using the synthetic mask , the three color components R, G, and B of the original image And the corrected color component images Perform image fusion to obtain a fused image , merge color channels and complete spatial target image enhancement.

2. The method for enhancing space target images under low light intensity according to claim 1, characterized in that: Step 7 is as follows: Decompose the original image into three color components R, G, and B to obtain the corresponding color component image , and adaptively derive the parameters according to the average gray value of each color component , as shown below: ； in, are the height and width of the original image respectively; Then through the adaptive gamma correction function For color component images Correction is performed to obtain the corrected color component image , as shown below: ； 。 3. The method for enhancing space target images under low light intensity according to claim 2, characterized in that: Step 8 is as follows: Step 8.

1. Detect the complete target area binary image The boundary of the target area is processed by Gaussian blur to obtain the blurred boundary image of the complete target area. ; Step 8.

2. Get the synthetic mask by the following formula : 。 4. The method for enhancing space target images under low light intensity according to any one of claims 1 to 3, characterized in that: In step 9, use , as well as Perform image fusion to obtain a fused image , specifically: 。 5. The method for enhancing space target images under low light intensity according to claim 4, characterized in that: In step 2, the threshold segmentation is specifically as follows: Luminance component map, according to The mean value of the brightness component image is used for threshold segmentation; In step 3, the edge detection adopts Canny edge detection; the isolated noise points in the binary result image are filtered out by median filtering; Step 4 is as follows: create an ellipse kernel with a radius of 19-22 pixels as a structural element; and then use the ellipse kernel to perform closing and opening morphological operations on the binary result image of the noise filtered out.

6. The method for enhancing space target images under low light intensity according to claim 5, characterized in that: In step 3, the merging method is phase OR; In step 4, an ellipse kernel with a radius of 20 pixels is created as a structural element.

7. A computer program product comprising a computer program, characterized in that: When the program is executed by a processor, the steps of the method for enhancing the image of a space target under low light intensity described in any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method for enhancing the image of a space target under low light intensity as claimed in any one of claims 1 to 6 are implemented.

Images