Multi-scale novel Top-Hat infrared small target detection method

Abstract

The invention discloses a novel multi-scale Top-Hat infrared small target detection method. The method comprises the following steps: step S110, obtaining an original image matrix containing an infrared small target; step S120, respectively constructing annular structure elements under different scales; step S130, performing multi-scale Top-Hat transformation on the original image matrix by using the constructed annular structure elements to generate enhanced target matrixes under multiple scales; step S140, calculating a non-zero mean value of each enhanced target matrix, and screening out the enhanced target matrix with the maximum mean value; s150, in the enhanced target matrix with the maximum mean value, extracting a part higher than the mean value as a suspected target area; and S160, calculating each connected domain in the suspected target area, and screening the connected domains according to a preset small target pixel number, thereby realizing target detection. According to the technical scheme, compared with structural elements of a single scale, targets of different sizes can be more effectively captured.

Description

Technical Field

[0001] The invention relates to the technical fields of image processing, target detection and image segmentation, and in particular to a novel multi-scale Top-Hat infrared small target detection method. Background Art

[0002] Multi-scale annular structure elements have important application value in image processing, especially in infrared small target detection. By introducing the concept of multi-scale, it can be more flexibly adapted to targets of different sizes and shapes. Multi-scale annular structure elements can capture the characteristics of the target at different scales, thereby effectively separating the target area from the complex background. This method can enhance the contrast of the target while reducing the interference of background noise and improving the accuracy of detection.

[0003] The Top-Hat bandpass filter is a powerful image processing tool, especially suitable for the detection of small targets. By using this filter, local features in the image can be effectively extracted while suppressing unnecessary background information. The Top-Hat transform can identify target areas that are brighter than the background by performing morphological operations on the image. Combined with multi-scale annular structural elements, the Top-Hat bandpass filter processes the image at different scales, and finally accurately screens out the real target based on the target characteristics, thereby improving the sensitivity and accuracy of infrared small target detection.

[0004] These technologies have their own shortcomings, mainly in terms of sensitivity to noise and complex backgrounds, and are prone to false detection and missed detection. The inventors note that there is currently no technology that can combine the advantages of both technologies to effectively avoid these shortcomings. Summary of the invention

[0005] The purpose of the present invention is to provide a novel multi-scale Top-Hat infrared small target detection method, which can enhance the target and improve the target detection rate in the infrared small target detection in complex background.

[0006] The technical solution adopted by the present invention is: a new multi-scale Top-Hat infrared small target detection method, comprising: Step S110, obtaining an original image matrix containing a small infrared target; Step S120, constructing ring structure elements at different scales respectively; Step S130, using the constructed annular structure element to perform a multi-scale Top-Hat transformation on the original image matrix to generate an enhanced target matrix at multiple scales; Step S140, calculating the non-zero mean of each enhanced target matrix, and selecting the enhanced target matrix with the maximum mean; Step S150, in the enhanced target matrix with the maximum mean value, extracting a portion higher than the mean value as a suspected target region; Step S160, calculating each connected domain in the suspected target area, and screening the connected domain according to a preset number of small target pixels, thereby achieving target detection.

[0007] The beneficial effects of the present invention compared with the prior art are: (1) By constructing ring-shaped structural elements for different scales, the algorithm can adapt to targets of various sizes and enhance the ability of multi-scale target detection. Compared with the structural elements of a single scale, it can more effectively capture targets of different sizes.

[0008] (2) The present invention uses the mean of non-zero pixels to screen out the most suitable scale. This statistical property-based screening method can automatically select the optimal scale and improve the accuracy of target detection. Compared with the traditional fixed threshold method, this dynamic selection method is more robust.

[0009] (3) The present invention determines the suspected target area through connected domain analysis and makes comparisons within the suspected target area, which can not only effectively segment the target, but also further confirm the authenticity of the target through the grayscale difference of the suspected target area, thereby reducing the false detection rate. This multi-stage screening and confirmation mechanism improves the accuracy of target detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a flow chart of the novel multi-scale Top-Hat infrared small target detection method of the present invention; Figure 2 The detection results of different algorithms, (a) is the original image, (b) is the result of LCM algorithm processing, (c) is the result of traditional Top-Hat algorithm processing, (d) is the result of RPCA algorithm processing, (e) is the result of this method processing; Figure 3 SCRG values ​​of different algorithms for each image, where (a) shows the SCRG values ​​of each algorithm for Figure 2 (a) shows the detection results of the top image, and (b) shows the detection results of each algorithm. Figure 2 (a) shows the detection results of the middle image, and (c) shows the detection results of each algorithm. Figure 2 (a) Detection results of the bottom image. DETAILED DESCRIPTION

[0011] The present invention is further described below with reference to the accompanying drawings and specific embodiments.

[0012] Small targets have the characteristics of small area and irregular shape, and infrared images have the characteristics of low contrast and high noise. Therefore, the signal-to-noise ratio of small targets in infrared images is low, especially in complex backgrounds. It is difficult for traditional Top-Hat methods to detect infrared small targets. In order to solve this problem, the present invention proposes a new multi-scale Top-Hat infrared small target detection method. By constructing a Top-Hat bandpass filter with multi-scale annular structure elements, a multi-scale Top-Hat transform is performed on the same infrared small target image, and finally the real target is screened out according to the target characteristics.

[0013] Figure 1 FIG. 1 is a flow chart of the novel multi-scale Top-Hat infrared small target detection method of the present invention. Figure 1 As shown, the method comprises the following steps: Step S110, obtaining an original image matrix containing a small infrared target; Step S120, constructing ring structure elements at different scales respectively; Constructing annular structure elements at different scales includes: first, determining multiple scale parameters, each scale parameter corresponds to an annular structure element. The size and thickness of the annular structure element are adjusted according to the preset scale parameters to adapt to targets of different sizes. By adjusting the inner and outer diameters of the annular structure element, annular structure elements of multiple scales can be generated. This ensures that target features at different scales can be more comprehensively captured in subsequent processing. Annular structure elements of different sizes are used for image processing at different scales.

[0014] Step S130, using the constructed annular structure element to perform a multi-scale Top-Hat transformation on the original image matrix to generate an enhanced target matrix at multiple scales; For each ring structure element, a Top-Hat transformation is defined. The Top-Hat transformation can be expressed by the following formula: , in is the ring structure element of the i-th scale, I is the original image matrix containing the infrared small target, Represents the result of opening I (erosion followed by expansion).

[0015] For each scale i, use the corresponding ring structure element Perform Top-Hat transformation on the original image matrix I to obtain the enhanced target matrix : , This process generates multiple enhanced target matrices , ,…, , each enhanced target matrix corresponds to a different scale.

[0016] Step S140, calculating the non-zero mean of each enhanced target matrix, and selecting the enhanced target matrix with the maximum mean; For each scale i, calculate the corresponding enhanced target matrix The non-zero mean of .exist , ,…, Find the maximum value in : , according to Corresponding scale, find the enhanced target matrix with the maximum mean .

[0017] Step S150, in the enhanced target matrix with the maximum mean value, extracting a portion higher than the mean value as a suspected target region; For the augmented target matrix with maximum mean , traverse each pixel, if the gray value of the pixel is less than or equal to , then set it to 0, otherwise, if it is greater than , then it is determined that the pixel is located in the suspected target area.

[0018] Step S160, calculating each connected domain in the suspected target area, and screening the connected domain according to a preset number of small target pixels, thereby achieving target detection.

[0019] Calculate each connected domain in the suspected target area, including determining its boundaries and counting the number of pixels. The upper limit of the number of pixels of small targets is preset to 25, and only the connected domains with less than 25 pixels are retained to filter out the enhanced image matrix containing the real target.

[0020] According to the technical solution of the present invention, combined with multi-scale annular structure elements, the Top-Hat bandpass filter processes the image at different scales, and finally accurately screens out the real target according to the target characteristics, thereby improving the sensitivity and accuracy of infrared small target detection.

[0021] In order to prove the effectiveness and detection ability of this method, three infrared small target images in complex backgrounds were selected for verification and compared with other commonly used target detection methods. The other three comparison algorithms are: Local Contrast Measure (LCM), traditional Top-Hat filtering algorithm and RPCA algorithm.

[0022] The test comparison results are as follows Figure 2As shown in the figure, (a) is the original image, (b) is the result of LCM algorithm processing, (c) is the result of traditional Top-Hat algorithm processing, (d) is the result of RPCA algorithm processing, and (e) is the result of this method processing.

[0023] In order to quantitatively compare the advantages and disadvantages of various algorithms, the signal-to-clutter ratio (SCR) and signal-to-clutter ratio gain (SCRG) are introduced: , , in, is the mean gray value of the target area, is the mean gray value of the background area, is the standard deviation of the grayscale value of the background area, Represents the clutter signal ratio of the original image, It represents the clutter signal ratio of the output image after algorithm processing. SCRG represents the enhancement effect of the algorithm on the target. The larger the SCRG value, the better the detection effect of the algorithm.

[0024] Figure 3 The SCRG values ​​in the figure show the effects of each algorithm on Figure 2 (a) shows the detection results of three original images, where (a) shows the detection results of each algorithm. Figure 2 (a) shows the detection results of the top image, and (b) shows the detection results of each algorithm. Figure 2 (a) shows the detection results of the middle image, and (c) shows the detection results of each algorithm. Figure 2 The detection result of the bottom image in (a). Figure 2 and Figure 3 , which proves that this method has good detection capability for small infrared targets in complex backgrounds.

[0025] The above is only a specific implementation of the present invention, but the protection scope of the present invention is not limited thereto. Any person familiar with the technology can understand and think of any changes or substitutions within the technical scope disclosed by the present invention, which should be included in the scope of the present invention. The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

Claims

1. A novel multi-scale Top-Hat infrared small target detection method, characterized in that: include: Step S110, obtaining an original image matrix containing a small infrared target; Step S120, constructing ring structure elements at different scales respectively; Step S130, using the constructed annular structure element to perform a multi-scale Top-Hat transformation on the original image matrix to generate an enhanced target matrix at multiple scales; Step S140, calculating the non-zero mean of each enhanced target matrix, and selecting the enhanced target matrix with the maximum mean; Step S150, in the enhanced target matrix with the maximum mean value, extracting a portion higher than the mean value as a suspected target region; Step S160, calculating each connected domain in the suspected target area, and screening the connected domain according to a preset number of small target pixels, thereby achieving target detection.

2. The novel multi-scale Top-Hat infrared small target detection method according to claim 1, characterized in that: In step S110, the area of ​​the small infrared target is not larger than 25 pixels.

3. The novel multi-scale Top-Hat infrared small target detection method according to claim 1, characterized in that: Step S130 includes: For each ring structure element, define the Top-Hat transformation: ， in, is the ring structure element of the i-th scale, I is the original image matrix containing the infrared small target, represents the result of opening I; For each scale i, use the corresponding ring structure element Perform Top-Hat transformation on the original image matrix I to obtain the enhanced target matrix : ， Among them, each enhanced target matrix corresponds to a different scale.

4. The novel multi-scale Top-Hat infrared small target detection method according to claim 1, characterized in that: In step S150, for the enhanced target matrix with the maximum mean, each pixel is traversed, and if the grayscale value of the pixel is less than or equal to the maximum mean, it is set to 0; otherwise, if it is greater than the maximum mean, the pixel is determined to be located in the suspected target area.

5. The novel multi-scale Top-Hat infrared small target detection method according to claim 1, characterized in that: Step S160 includes: marking and calculating the number of pixels of each connected domain in the suspected target area, and only retaining the connected domains whose number of pixels is less than the preset number of small target pixels, thereby screening and obtaining an enhanced image matrix containing the real target.

Images