Optical density reconstructed plankton microscopic image enhancement method and system

Abstract

The application discloses a kind of plankton microscopic image enhancement method and system of optical density reconstruction, belong to resource and environmental technical field.Firstly, the microscopic image is converted to Lab space;In L channel, optical density map is calculated according to Lambert-Beer law, the highest M point mean value of brightness is taken as incident light intensity I0, the texture is enhanced by reconstructing L channel initial map, Gamma transformation and local histogram equalization;Local variance is calculated in a, b channel, color retention weight is obtained by Tanh mapping, and color cast is corrected in combination with background color mean value.Finally, the enhanced Lab image is converted back to RGB format.Through testing, the method can effectively solve the problem of blurred limb features caused by lugol reagent fixation, and provide key technical support for plankton species identification.

Description

Technical Field

[0001] This invention belongs to the field of resource and environmental technology, and specifically relates to a method and system for enhancing zooplankton microscopic images by optical density reconstruction. Background Technology

[0002] Zooplankton are indicator organisms of aquatic ecosystems, and their species distribution is an important indicator for assessing aquatic ecological environments. Currently, zooplankton identification still mainly relies on "field sampling and laboratory analysis." To preserve natural water samples for long periods, zooplankton need to be fixed using Lugol's reagent. Zooplankton themselves are colorless or even transparent, but after fixation with Lugol's reagent, they are stained a deep yellow, making key limb features indistinct or even lost, thus seriously interfering with zooplankton species identification.

[0003] Image enhancement is a technique that utilizes digital image processing technology to improve the recognizability of images. It mainly consists of two major technical approaches: non-physical model methods and physical model methods. Non-physical model methods primarily include digital image processing techniques such as histogram equalization and Retinex, as well as deep learning-based image enhancement methods. Physical model methods utilize the properties of light propagation and absorption to construct image enhancement methods, such as dark channel prior image dehazing algorithms based on atmospheric scattering models. Existing methods have been widely applied in macroscopic fields such as image dehazing; however, there are currently no reports on how to enhance zooplankton microscopic images fixed with Lugol reagent to improve the resolution of limb details in these images. Summary of the Invention

[0004] To address the problem that key limb features in zooplankton microscopic images are not obvious or even lost due to the use of Lugo's reagent to fix water samples, this invention discloses a method and system for enhancing zooplankton microscopic images by optical density reconstruction. This method reduces the impact of Lugo's reagent on the quality of zooplankton microscopic images, improves the distinguishability of limb details, and provides key technical support for zooplankton species identification.

[0005] The technical solution of the present invention is as follows:

[0006] A method for enhancing zooplankton microscopic images through optical density reconstruction includes the following steps:

[0007] Convert the microscopic images of zooplankton fixed with Lugo's reagent to the Lab color space;

[0008] L-channel reconstruction: Calculate the optical density map of the L-channel, take the average value of the M pixels with the highest brightness as the incident light intensity I0, and reconstruct the initial L-channel map using the formula;

[0009] Texture enhancement: Gamma transformation and adaptive local histogram equalization are performed on the normalized optical density map;

[0010] Color correction: Calculate the local variance of channels a and b, obtain the color retention weight through Tanh mapping, and correct channels a and b by combining the background color mean.

[0011] In the above technical solution, the L-channel optical density map is calculated using the following formula:

[0012] (1)

[0013] in, This is the optical density map for the L channel. These are the pixel coordinates. I0 is the brightness value of the L channel, and I0 is the incident light intensity, which is the average value of the M pixels with the highest brightness.

[0014] In the above technical solution, the formula for calculating the incident light intensity I0 is:

[0015] (2)

[0016] in, , Here, N is the total number of pixels. It is the i-th largest brightness value after sorting.

[0017] In the above technical solution, in the calculation formula of I0, .

[0018] In the above technical solution, the initial image reconstruction formula for the L channel is:

[0019] (3)

[0020] For the initial image of the L channel, max(D) c () represents the maximum value of the L-channel optical density map. I 0 represents the intensity of the incident light.

[0021] In the above technical solution, the local variances of channels a and b are calculated using the following formula:

[0022] (4)

[0023] in, Let a and b be the local variances. for , variance These are the a and b channels of the original image in the Lab color space. The window size represents the local variance. These are linear coefficients.

[0024] In the above technical solution, the color retention weight is obtained through an inversion operation, and the corrected color channel... and They are respectively:

[0025] (5)

[0026] (6)

[0027] in, Let a be the weight. Let b be the weight. The background color of a Set the background color for b.

[0028] In the above technical solution, the local variance window size linear coefficients .

[0029] A zooplankton species identification system integrating the above methods, comprising:

[0030] Microscopic image acquisition module;

[0031] Lab color space conversion module;

[0032] Optical density reconstruction processor.

[0033] A computer-readable storage medium storing program instructions that, when executed by a processor, implement the above-described method.

[0034] Beneficial effects:

[0035] 1. Breakthrough in feature reconstruction:

[0036] It overcomes the masking effect of Lugo reagent staining, improving the clarity of key identification features such as zooplankton appendages and bristles; the PSNR index surpasses that of traditional methods.

[0037] 2. Color correction accuracy:

[0038] The color shift correction mechanism, which combines local variance with Tanh mapping, preserves the authenticity of background colors; it avoids the drawback of Dehaze-type methods that amplify background noise and focuses more on the target area.

[0039] 3. Improved identification efficiency:

[0040] Species identification accuracy is improved; processing time per image is reduced, meeting the needs of real-time analysis. Attached Figure Description

[0041] Figure 1 This is a technical roadmap for a method of enhancing zooplankton microscopic images by optical density reconstruction according to the present invention.

[0042] Figure 2 A comparison of different methods for enhancing zooplankton microscopic images. Detailed Implementation

[0043] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. However, the following embodiments are only for explaining the present invention, and the scope of protection of the present invention should include all the contents of the claims. Moreover, through the description of the following embodiments, those skilled in the art can fully implement all the contents of the claims of the present invention.

[0044] Example

[0045] This embodiment discloses a method and system for enhancing zooplankton microscopic images using optical density reconstruction. The method first converts the zooplankton microscopic image to the Lab color space; then, it reconstructs the zooplankton microscopic image using optical density maps in the L channel, and performs color restoration using local variance in the a and b channels. This reduces the impact of Lugol reagent fixation on the quality of the zooplankton microscopic image and improves the resolution of detailed features of zooplankton limbs. The technical route of this invention is as follows: Figure 1 As shown, the specific, simplified implementation steps are as follows:

[0046] (1) Calculate the optical density map of the L channel of the zooplankton micrograph.

[0047] According to Beer-Lambert law, the incident light intensity I0 attenuates as it passes through the microscopic field of view of zooplankton, and the optical density of each pixel received by the CCD camera... As shown in formula (1).

[0048] (1)

[0049] in, The L channel represents the zooplankton microscopic image received by the CCD camera. This method acquires the pixel coordinates of an image from a CCD camera. Zooplankton microscopic images consist of two parts: the zooplankton target area and the background area. The background area is unobstructed by the target, and its L-channel brightness value is relatively transparent. This invention assumes that the incident light intensity I0 within a microscopic field of view is stable, and takes the average of the M pixels with the highest L-channel brightness values ​​as the incident light intensity I0, i.e.:

[0050] (2)

[0051] in, N is the total number of pixels. It is the i-th largest brightness value after sorting. In this invention's example... .

[0052] Then, by inverting the normalized optical density map, the reconstructed initial L-channel map is obtained. As shown in formula (3):

[0053] (3)

[0054] (2) Enhance texture detail features

[0055] After normalizing the L-channel optical density map, the initial map is enhanced by utilizing nonlinear variations of Gamma. The contrast of severely stained areas was improved, and then adaptive local histograms were used to enhance the texture details of the zooplankton micrographs, resulting in enhanced channels. .

[0056] (3) Color correction of channel a and channel b

[0057] First, the local variances of color channels a and b are calculated. Then, the local variance values ​​are mapped to the [0,1] interval using the Tanh transform, and the weights for retaining the background color are obtained by inverting the values. .

[0058] (4)

[0059] in, Let a and b be the local variances. for , variance These are the a and b channels of the original image in the Lab color space. The window size represents the local variance. These are linear coefficients. In this invention's example, the window size for local variance is... , .

[0060] The background color is determined by taking the average values ​​of the a and b channels corresponding to the M pixels with the highest brightness values ​​in the L channel. The corrected color channels and They are respectively:

[0061] (5)

[0062] (6)

[0063] Finally, the Lab format of the zooplankton microscopic images enhanced by this invention is... You can obtain enhanced zooplankton microscopic images by converting them to RGB format.

[0064] Experimental Results Analysis and Discussion:

[0065] Figure 2 Visual comparison results are presented between this application and four typical image enhancement methods: SSR, MSR, MSRCR (Multi-scale Retina Enhancement Algorithm), and Dehazed. In the original image, due to Lugol's staining, the limb details in the zooplankton microscopic image are severely obscured. The visual comparison results show that this invention has significant advantages, allowing for the observation of much clearer zooplankton limb details.

[0066] Table 1 objectively compares the present invention with the comparative methods from four perspectives: Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), Entropy, and Gradient. The objective evaluation results show that the present invention has significant advantages in terms of PSNR and SSIM. Although the results of the present invention are lower than Dehazed's in terms of Entropy and Gradient, combined with… Figure 2 Subjective evaluations show that the Dehazed enhancement method did not significantly enhance the detailed texture of the zooplankton target region. Instead, it amplified background noise in the zooplankton microscopic image, thus increasing information entropy and gradient values. In summary, compared with the comparative methods, this invention has significant advantages. It effectively solves the problem of zooplankton limb detail features being masked by Lugol's reagent staining, significantly enhances the zooplankton limb detail features in the microscopic image, and provides an effective image enhancement method for zooplankton species identification.

[0067] Table 1

[0068]

[0069] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for enhancing zooplankton microscopic images using optical density reconstruction, characterized in that, include: Convert the microscopic images of zooplankton fixed with Lugo's reagent to the Lab color space; L-channel reconstruction: The optical density map of the L-channel is calculated based on the Lambert-Beer law. The average value of the M pixels with the highest brightness is taken as the incident light intensity I0. The initial L-channel map is then reconstructed using the formula. Texture enhancement: Gamma transformation is applied to the normalized optical density map to enhance the contrast of severely stained areas, and adaptive local histogram equalization is performed; Color correction: Calculate the local variance of channels a and b, obtain the color retention weight through Tanh mapping, and correct channels a and b by combining the background color mean; The L-channel optical density map is calculated using the following formula: (1) in, These are the pixel coordinates. I0 is the luminance value of the L channel and I0 is the incident light intensity. The formula for calculating the incident light intensity I0 is as follows: (2) in, , Here, N is the total number of pixels. It is the i-th largest brightness value after sorting, in the formula for calculating I0. ; The initial image reconstruction formula for the L-channel is: (3) For the initial image of the L channel, max(D) c )for The maximum value of I0 is the incident light intensity. The local variances of channels a and b are calculated using the following formula: (4) in, Let a and b be the local variances. for , variance These are the a and b channels of the original image in the Lab color space. The window size represents the local variance. The coefficients are linear. The color preservation weights are obtained through an inversion operation, resulting in corrected color channels. and They are respectively: (5) (6) in, Let a be the weight. Let b be the weight. The background color of a Set the background color for b.

2. The method according to claim 1, characterized in that: The local variance window size linear coefficients .

3. A zooplankton species identification system, characterized in that: Integrating the method of claim 1 or 2, comprising: Microscopic image acquisition module; Lab color space conversion module; Optical density reconstruction processor.

4. A computer-readable storage medium storing program instructions, characterized in that: When the program instructions are executed by the processor, they implement the method of claim 1 or 2.

Images