Image cognitive interpretation information density evaluation and promotion method

Abstract

The application relates to an image cognitive interpretation information density evaluation and promotion method, and belongs to the technical field of optical imaging application. The method is as follows: starting from the human eye cognitive mechanism and interpretation process, image interpretation quality influence elements and key parameters and the correlation between the two are extracted; the cognitive interpretation information density evaluation model is established by comprehensively considering the key parameters; based on the cognitive interpretation information density evaluation model, a processing algorithm is designed to realize fine promotion of image interpretation quality by taking into account the transmission function promotion, texture detail preservation and noise artifact suppression. The application solves the problem that the interpretation quality elements are not clear in the prior art, fills the technical gap of effective information content quantitative evaluation, breaks through the technical bottleneck that the traditional processing is easy to cause artifacts, forms a complete technical system from information density evaluation to interpretation quality promotion, improves the ground object classification precision and target recognition rate, and provides key support for intelligent application of optical remote sensing images.

Description

Technical Field

[0001] This invention relates to a method for evaluating and improving information density in image cognitive interpretation, belonging to the field of optical imaging application technology. Background Technology

[0002] In the fields of aviation and aerospace, various optical remote sensing images, with their ability to quickly and accurately acquire information about ground features, have become an indispensable technical means for applications such as nature monitoring, geographic exploration, and environmental surveillance. However, the optical remote sensing imaging process is constrained by multiple inherent factors such as atmospheric interference, system aberrations, sensor noise, and geometric radiation distortion, resulting in problems such as hazy images, detail degradation, and noise artifacts, which severely limit the effective extraction and application depth of ground feature information.

[0003] With the surge in demand for high-resolution, high-time-sensitivity Earth observation, fields such as detailed land monitoring, disaster emergency response, environmental dynamic assessment, and military target identification have placed unprecedentedly high standards on the clarity, accuracy, fidelity, and reliability of image data. Currently, the optical remote sensing imaging chain is complex, with highly coupled influencing factors. Existing technologies lack effective methods and enhancement techniques for evaluating information content in human-in-the-loop interpretation applications, leading to problems such as noise amplification and artifact formation in processed images. Improving and ensuring the quality of original images has become a core technical bottleneck restricting accurate interpretation.

[0004] Improving the quality of optical remote sensing image interpretation is a key prerequisite for unlocking the value of remote sensing big data, directly impacting the effectiveness of major strategic tasks such as global change research, resource management, and national security. While existing technologies have made some progress in areas such as image restoration and super-resolution reconstruction, a complete technical system has not yet been formed, encompassing everything from quantitative evaluation of information density to targeted improvement of interpretation quality, making it difficult to meet the urgent need to enhance the application efficiency of optical satellite systems.

[0005] High-quality interpretation images can significantly improve the accuracy of ground feature classification, target recognition rate, and change detection sensitivity, and are a core support for promoting the transformation and upgrading of remote sensing applications from qualitative to quantitative, from static to dynamic, and from manual to intelligent. Breaking through the technical bottlenecks in evaluating and improving the effective information density of optical remote sensing images is not only related to the usability and reliability of remote sensing information products, but also a strategic requirement for enhancing the country's independent spatial information security capabilities and seizing the commanding heights of remote sensing technology. Summary of the Invention

[0006] To address the problems existing in the background technology, the present invention provides a method for evaluating and improving the information density of image cognitive interpretation.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a method for evaluating and improving information density in image cognitive interpretation, the method comprising the following steps:

[0008] S1: Starting from the human eye's cognitive mechanism and interpretation process, extract the factors and key parameters that affect the quality of image interpretation and sort out the relationship between them;

[0009] S2: Taking into account key parameters, establish a cognitive interpretation information density evaluation model;

[0010] S3: Based on the cognitive interpretation information density evaluation model, with the goal of balancing transfer function improvement, texture detail preservation, and noise artifact suppression, a processing algorithm is designed to achieve a fine improvement in image interpretation quality.

[0011] Furthermore, the image interpretation quality influencing factors mentioned in S1 include the target's shape, texture, scale, shadows, and activity features.

[0012] Furthermore, the key parameters mentioned in S1 include ground resolution, dynamic range, transfer function, ringing aliasing, and signal-to-noise ratio.

[0013] Furthermore, the transfer function is the on-orbit end-to-end normalized transfer function H of the optical remote sensing satellite:

[0014] (1)

[0015] In formula (1):

[0016] This represents the contribution of the transfer function of the optical system;

[0017] The transfer function contribution of atmospheric transport is represented.

[0018] This represents the contribution of the detector's transfer function;

[0019] The transfer function contribution represents the motion of the satellite platform;

[0020] The cutoff frequency of the transfer function is , This refers to the pixel size.

[0021] Furthermore, the dynamic range refers to the effective grayscale number of the image, which is the total number of grayscale levels in the image where the proportion of grayscale pixels exceeds a threshold. The effective quantization formula for the effective grayscale number L is as follows:

[0022] (2)

[0023] In formula (2):

[0024] This indicates the effective dynamics of the image.

[0025] Furthermore, the signal-to-noise ratio includes the imaging signal-to-noise ratio and the quantization signal-to-noise ratio;

[0026] The imaging signal-to-noise ratio for:

[0027] (3)

[0028] In formula (3):

[0029] Indicates the standard deviation of the signal;

[0030] Indicates the standard deviation of imaging noise;

[0031] Indicates the average number of electrons in the imaging;

[0032] Indicates device noise;

[0033] Indicates circuit noise;

[0034] Indicates charge conversion efficiency;

[0035] The quantization signal-to-noise ratio for:

[0036] (4)

[0037] In equation (4):

[0038] K represents the gain;

[0039] Indicates the standard deviation of quantization noise;

[0040] c represents the quantization interval width adjustment parameter.

[0041] Furthermore, the ringing artifacts are distributed in the high-frequency aliasing region, as shown below:

[0042] (5)

[0043] In equation (5):

[0044] ALI represents the power spectrum of aliasing ringing information;

[0045] This represents the ringing intensity parameter;

[0046] Represents the normalized power spectrum of the physical scene;

[0047] Represents bidirectional frequency coordinates in the frequency domain;

[0048] Indicates the aliasing makeup function:

[0049] (6)

[0050] In equation (6):

[0051] Represents the impact function;

[0052] Indicates the horizontal sampling interval;

[0053] Indicates the vertical sampling interval;

[0054] m and n both represent the summation count number.

[0055] Furthermore, the cognitive interpretation information density evaluation model described in S2 is as follows:

[0056] (7)

[0057] In equation (7):

[0058] This represents the information density of cognitive interpretation;

[0059] Indicates the focal length of the imaging system;

[0060] Indicates the imaging distance;

[0061] Indicates the physical size of a pixel in the horizontal direction;

[0062] Indicates the physical size of the pixel in the vertical direction;

[0063] This symbol represents a transfer and has no physical meaning.

[0064] Indicates the number of spectral channels in the image;

[0065] Indicates the spectral adjustment parameter;

[0066] This indicates that the parameters have been effectively quantified and adjusted.

[0067] Furthermore, the processing algorithm described in S3 is an adaptive prior regularization method, and the optimized model is:

[0068] (8)

[0069] In equation (8):

[0070] This represents the original high-resolution image.

[0071] Both p and p represent prior fitting parameters for scene detail fidelity, which are optimization variables;

[0072] Represents the point spread function;

[0073] This indicates minimization optimization.

[0074] Represents the regularization parameter matrix;

[0075] This indicates a degraded image from the imaging system;

[0076] This represents the smooth 0-norm.

[0077] Compared with the prior art, the beneficial effects of the present invention are:

[0078] This invention, starting from the cognitive mechanism and interpretation process of the human eye, systematically sorts out the correlation between the factors affecting image interpretation quality and key parameters, solving the problem of unclear interpretation quality factors in existing technologies; it constructs a cognitive interpretation information density evaluation model, filling the technical gap in the quantitative evaluation of effective information content; it designs a processing algorithm to achieve coordinated control of transfer function optimization, detail preservation and noise suppression, breaking through the technical bottleneck of traditional processing easily causing artifacts; it forms a complete technical system from information density evaluation to interpretation quality improvement, improving the accuracy of ground object classification and target recognition rate, and providing key support for the intelligent application of optical remote sensing images. Attached Figure Description

[0079] Figure 1 This is a schematic diagram of the decomposition of the cognitive interpretation influence parameters of the present invention;

[0080] Figure 2 This is a diagram illustrating the effect of improved interpretation quality. Detailed Implementation

[0081] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0082] A method for evaluating and enhancing information density in image cognitive interpretation is a method for evaluating and enhancing information density in remote sensing images for applications of human visual interpretation and machine autonomous interpretation. The method includes the following steps:

[0083] S1: When interpreting high-value targets in an image, starting from the human eye's cognitive mechanism and interpretation process, extract the factors and key parameters that affect the quality of image interpretation and sort out the relationship between them;

[0084] Interpretation quality is not solely determined by the image's inherent information quality (such as sharpness, noise, and other basic attributes), but is a comprehensive concept that integrates multiple dimensions. Specifically, it includes:

[0085] Image features: such as the objective visual attributes of the target, including its shape, texture, and scale;

[0086] Visual perception: The subjective feeling and recognition rules of the human eye towards images;

[0087] Professional knowledge: The interpreter's judgment based on domain experience (such as geology, military, etc.);

[0088] Analysis and computation: processing and quantifying image data through algorithms;

[0089] Intelligence interpretation: In-depth mining and interpretation of image information in conjunction with mission objectives.

[0090] This means that when evaluating the quality of interpretation, we need to move beyond focusing on the single physical attributes of an image and take into account the combination of subjective perception and objective data, as well as basic features and application goals.

[0091] An image information density evaluation model needs to be constructed, and the construction of the evaluation model should follow the principle of "prior information + actual needs + human eye guidance":

[0092] Regarding the foundation of prior information: Based on the physical light field (such as the laws of light propagation) and imaging system parameters (such as optical system, detector performance, etc.), the physical characteristics of the image itself are clearly defined;

[0093] Regarding application requirements: different scenarios (such as disaster monitoring and military identification) have different focuses on images, and the model needs to be adapted to specific task objectives;

[0094] Regarding human eye interpretation guidance: Ultimately, it serves the human eye's cognition and interpretation of images, ensuring that the model results conform to the laws of human visual understanding.

[0095] The key parameters in the model represent the effectiveness of image information from different dimensions:

[0096] Transfer function: determines the completeness of scene information representation, that is, whether the image can completely convey the detailed features of the ground objects (such as edges and textures). The better the performance of the transfer function, the less information is lost.

[0097] Dynamic range and effective quantization: These determine the completeness of the scene's dynamic information acquisition. Dynamic range reflects the image's ability to present differences in brightness and darkness, while effective quantization reflects the efficiency of grayscale utilization. Together, they ensure the capture of information from complex scenes (such as areas with alternating brightness and darkness).

[0098] Noise artifact suppression capability: characterizes the richness of the original scene information retained by the image. Noise and artifacts (such as the ringing effect generated during processing) will interfere with the real information. The stronger the suppression capability, the higher the fidelity of the original ground feature information.

[0099] S2: Based on information theory and optical remote sensing imaging mechanism, the existing information density characterization model is extended to the physical domain. Taking into account key parameters, a cognitive interpretation information density evaluation model is established to provide theoretical support for improving image interpretation quality.

[0100] S3: Based on the cognitive interpretation information density evaluation model, it can be seen that the improvement of ground image quality is a multi-objective comprehensive processing improvement problem. When the resolution remains unchanged, the goal is to improve the transfer function, preserve texture details, and suppress noise artifacts. The processing algorithm is designed to achieve a fine improvement in image interpretation quality, so as to enhance the application capabilities of optical satellite images.

[0101] Furthermore, the image interpretation quality influencing factors mentioned in S1 include the target's shape, texture, scale, shadows, and activity features.

[0102] Furthermore, the key parameters mentioned in S1 include ground resolution, dynamic range, transfer function, ringing aliasing, and signal-to-noise ratio.

[0103] Furthermore, the transfer function is the on-orbit end-to-end normalized transfer function H (maximum value normalized) of the optical remote sensing satellite:

[0104] (1)

[0105] In formula (1):

[0106] This represents the contribution of the transfer function of the optical system;

[0107] The transfer function contribution of atmospheric transport is represented.

[0108] This represents the contribution of the detector's transfer function;

[0109] The transfer function contribution represents the motion of the satellite platform;

[0110] The cutoff frequency of the transfer function needs to take into account the actual physical size of the pixel. , This refers to the pixel size.

[0111] Furthermore, the dynamic range refers to the effective grayscale number of the image, which is the total number of grayscale levels in the image where the proportion of grayscale pixels exceeds a threshold, where the threshold is 0.001%. The effective quantization L formula for the effective grayscale number of the image is as follows:

[0112] (2)

[0113] In formula (2):

[0114] This indicates the effective dynamics of the image.

[0115] Furthermore, the signal-to-noise ratio includes the imaging signal-to-noise ratio and the quantization signal-to-noise ratio;

[0116] The imaging signal-to-noise ratio (Unit is 1) is:

[0117] (3)

[0118] In formula (3):

[0119] Indicates the standard deviation of the signal;

[0120] Indicates the standard deviation of imaging noise;

[0121] It represents the average number of electrons in imaging, which is related to detector area, average quantum efficiency, spectral band, integration time, target radiance, optical transmittance, etc.

[0122] Indicates device noise;

[0123] Indicates circuit noise;

[0124] Indicates charge conversion efficiency;

[0125] The quantization signal-to-noise ratio (Unit is 1) is:

[0126] (4)

[0127] In equation (4):

[0128] K represents the gain;

[0129] Indicates the standard deviation of quantization noise;

[0130] c represents the quantization interval width adjustment parameter.

[0131] Furthermore, the ringing artifacts are distributed in the high-frequency aliasing region, as shown below:

[0132] (5)

[0133] In equation (5):

[0134] ALI represents the power spectrum of aliasing ringing information;

[0135] This parameter represents the ringing intensity, characterizing the distortion introduced by the processing algorithm, the algorithm's own principle, and the strength of the processed signal. related;

[0136] Represents the normalized power spectrum of the physical scene;

[0137] Represents bidirectional frequency coordinates in the frequency domain;

[0138] Indicates the aliasing makeup function:

[0139] (6)

[0140] In equation (6):

[0141] Represents the impact function;

[0142] Indicates the horizontal sampling interval;

[0143] Indicates the vertical sampling interval;

[0144] Both m and n represent summation counters and are positive integers.

[0145] Furthermore, the cognitive interpretation information density evaluation model described in S2 is as follows:

[0146] (7)

[0147] In equation (7):

[0148] This represents the information density of cognitive interpretation;

[0149] Indicates the focal length of the imaging system;

[0150] This indicates the imaging distance; when imaging at the nadir point, it is equal to the orbital altitude.

[0151] Indicates the physical size of a pixel in the horizontal direction;

[0152] Indicates the physical size of the pixel in the vertical direction;

[0153] This symbol represents a transfer and has no physical meaning.

[0154] Indicates the number of spectral channels in the image;

[0155] Indicates the spectral adjustment parameter;

[0156] This indicates that the parameters have been effectively quantified and adjusted.

[0157] and Characterize the impact of the number of spectral channels and effective quantization on the current interpretation task.

[0158] The cognitive interpretation information density evaluation model shows that: the higher the spatial resolution, the higher the effective quantization, and the higher the transfer function, the richer the interpretation information; the less aliasing and ringing, and the higher the signal-to-noise ratio, the higher the fidelity of the interpretation information and the stronger the image interpretation application capability, which can provide a theoretical basis for the optimization of imaging systems and the improvement of image interpretation quality.

[0159] Furthermore, the processing algorithm described in S3 is an adaptive prior regularization method, and the optimized model is:

[0160] (8)

[0161] In equation (8):

[0162] This represents the original high-resolution image.

[0163] Both p and p represent prior fitting parameters for scene detail fidelity, which are optimization variables;

[0164] Represents the point spread function;

[0165] This indicates minimization optimization.

[0166] Represents the regularization parameter matrix, and performs frequency band differentiation processing;

[0167] This indicates a degraded image from the imaging system;

[0168] This represents the smooth 0-norm.

[0169] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0170] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for evaluating and improving information density in image cognitive interpretation, characterized in that: The method includes the following steps: S1: Starting from the human eye's cognitive mechanism and interpretation process, extract the factors and key parameters that affect the quality of image interpretation and sort out the relationship between them; The factors affecting image interpretation quality include the target's shape, texture, scale, shadows, and motion features; The key parameters include ground resolution, dynamic range, transfer function, ringing aliasing, and signal-to-noise ratio; S2: Taking into account key parameters, establish a cognitive interpretation information density evaluation model; The cognitive interpretation information density evaluation model is as follows: (7) In equation (7): This represents the information density of cognitive interpretation; Indicates the focal length of the imaging system; Indicates the imaging distance; Indicates the physical size of a pixel in the horizontal direction; Indicates the physical size of the pixel in the vertical direction; This symbol represents a transfer and has no physical meaning. Indicates the number of spectral channels in the image; Indicates the spectral adjustment parameter; This indicates that the parameters have been effectively quantified and adjusted. L represents the effective quantization of the effective gray levels of the image; Represents the normalized power spectrum of the physical scene; Represents bidirectional frequency coordinates in the frequency domain; H represents the transfer function; Indicates the ringing intensity parameter; This represents the aliasing makeup function; Indicates the standard deviation of the signal; Indicates the standard deviation of imaging noise; K represents the gain; S3: Based on the cognitive interpretation information density evaluation model, an algorithm is designed to improve the quality of image interpretation by taking into account the improvement of transfer function, preservation of texture details and suppression of noise artifacts.

2. The method for evaluating and improving information density in image cognitive interpretation according to claim 1, characterized in that: The transfer function is the on-orbit end-to-end normalized transfer function H of the optical remote sensing satellite: (1) In formula (1): This represents the contribution of the transfer function of the optical system; The transfer function contribution of atmospheric transport is represented. This represents the contribution of the detector's transfer function; The transfer function contribution represents the motion of the satellite platform; The cutoff frequency of the transfer function is , This refers to the pixel size.

3. The method for evaluating and improving information density in image cognitive interpretation according to claim 2, characterized in that: The dynamic range refers to the effective grayscale number of the image, which is the total number of grayscale levels in the image where the proportion of grayscale pixels exceeds a threshold. The effective quantization formula for the effective grayscale number L is as follows: (2) In formula (2): This indicates the effective dynamics of the image.

4. The method for evaluating and improving information density in image cognitive interpretation according to claim 3, characterized in that: The signal-to-noise ratio includes the imaging signal-to-noise ratio and the quantization signal-to-noise ratio; The imaging signal-to-noise ratio for: (3) In formula (3): Indicates the standard deviation of the signal; Indicates the standard deviation of imaging noise; Indicates the average number of electrons in the imaging; Indicates device noise; Indicates circuit noise; Indicates charge conversion efficiency; The quantization signal-to-noise ratio for: (4) In equation (4): K represents the gain; Indicates the standard deviation of quantization noise; c represents the quantization interval width adjustment parameter.

5. The method for evaluating and improving information density in image cognitive interpretation according to claim 4, characterized in that: The ringing artifacts are distributed in the high-frequency aliasing region, as shown below: (5) In equation (5): ALI represents the power spectrum of aliasing ringing information; Indicates the ringing intensity parameter; Represents the normalized power spectrum of the physical scene; Represents bidirectional frequency coordinates in the frequency domain; Indicates the aliasing makeup function: (6) In formula (6): Represents the impulse function; Indicates the horizontal sampling interval; Indicates the vertical sampling interval; m and n both represent the summation count number.

6. The method for evaluating and improving information density in image cognitive interpretation according to claim 1 or 5, characterized in that: The processing algorithm described in S3 is an adaptive prior regularization method, and the optimized model is: (8) In equation (8): This represents the original high-resolution image. Both p and p represent prior fitting parameters for scene detail fidelity, which are optimization variables; Represents the point spread function; This indicates minimization optimization. Represents the regularization parameter matrix; This indicates a degraded image from the imaging system; This represents the smooth 0-norm.

Images