CN-120807484-B - Image cognition interpretation information density evaluation and promotion method

CN120807484BCN 120807484 BCN120807484 BCN 120807484BCN-120807484-B

Abstract

An image cognition interpretation information density evaluation and promotion method belongs to the technical field of optical imaging application. The method comprises the steps of starting from a human eye cognition mechanism and an interpretation flow, extracting an incidence relation between an image interpretation quality influence element and a key parameter, comprehensively considering the key parameter, establishing a cognition interpretation information density evaluation model, and designing a processing algorithm to achieve fine improvement of image interpretation quality based on the cognition interpretation information density evaluation model with the aim of considering transfer function improvement, texture detail maintenance and noise artifact suppression. The invention solves the problem of undefined interpretation quality factors in the prior art, fills the technical blank of quantitative evaluation of effective information content, breaks through the technical bottleneck that artifacts are easy to cause in the traditional processing, forms a complete technical system from information density evaluation to interpretation quality improvement, improves the classification precision of land features and the target recognition rate, and provides key support for intelligent application of optical remote sensing images.

Inventors

JIANG SHIKAI
ZHI XIYANG
HUANG YUANXIN
Bao Guangzhen
CHEN WENBIN
HU JIANMING
GONG JINNAN
ZHANG WEI

Assignees

哈尔滨工业大学

Dates

Publication Date: 20260512
Application Date: 20250808

Claims (6)

1. The image cognition interpretation information density evaluation and promotion method is characterized by comprising the following steps of: S1, extracting an image interpretation quality influence element and a key parameter from a human eye cognition mechanism and an interpretation flow, and combing the association relation of the image interpretation quality influence element and the key parameter; The image interpretation quality influence factors comprise the shape, texture, scale, shadow and activity characteristics of the target; The key parameters comprise ground resolution, dynamic range, transfer function, ringing aliasing and signal-to-noise ratio; s2, comprehensively considering key parameters and establishing a cognitive interpretation information density evaluation model; the cognitive interpretation information density evaluation model is as follows: (7) in the formula (7): representing the cognitive interpretation information density; Representing the focal length of the imaging system; Representing an imaging distance; representing the physical size of the pixels in the horizontal direction; representing the physical size of the pixels in the vertical direction; the representation is converted into a symbol, and has no physical meaning; representing the number of spectral channels of the image; representing a spectral adjustment parameter; Representing an effective quantization adjustment parameter; l represents the effective quantization of the effective gray scale number of the image; representing a normalized power spectrum of the physical scene; representing frequency domain bi-directional frequency coordinates; H represents a transfer function; representing a ringing intensity parameter; Representing an aliased vanity function; representing the standard deviation of the signal; Representing imaging noise standard deviation; K represents gain; and S3, based on the cognition interpretation information density evaluation model, aiming at considering transfer function improvement, texture detail maintenance and noise artifact suppression, designing a processing algorithm to improve the interpretation quality of the image.
2. The method for evaluating and promoting the image cognitive interpretation information density according to claim 1, wherein the transfer function is an optical remote sensing satellite on-orbit full-link normalization transfer function H: (1) In the formula (1): Representing the transfer function contribution of the optical system; representing the transfer function contribution of atmospheric transport; representing the transfer function contribution of the detector; a transfer function contribution representing satellite platform motion; The transfer function is cut off to the frequency , Is the pixel size.
3. The method for evaluating and promoting image recognition interpretation information density as claimed in claim 2, wherein the dynamic range is the effective gray scale number of the image, namely, the total number of gray scales with gray scale value pixels having a ratio exceeding a threshold in the image, and the effective quantization L of the effective gray scale number of the image has the following calculation formula: (2) In the formula (2): representing the effective dynamics of the image.
4. The method for evaluating and enhancing the density of image recognition interpretation information as claimed in claim 3, wherein the signal-to-noise ratio comprises an imaging signal-to-noise ratio and a quantization signal-to-noise ratio; the imaging signal to noise ratio The method comprises the following steps: (3) in the formula (3): representing the standard deviation of the signal; Representing imaging noise standard deviation; representing the imaging average electron number; Representing device noise; representing circuit noise; Representing charge conversion efficiency; the quantized signal to noise ratio The method comprises the following steps: (4) In the formula (4): K represents gain; Representing the standard deviation of quantization noise; c represents quantization interval width adjustment parameters.
5. The method for evaluating and enhancing image recognition interpretation information density as claimed in claim 4, wherein the ringing artifacts are distributed in a high-frequency aliasing area, expressed as: (5) in formula (5): ALI represents ALI ringing information power spectrum; representing a ringing intensity parameter; representing a normalized power spectrum of the physical scene; representing frequency domain bi-directional frequency coordinates; representing the aliased vanity function: (6) in formula (6): Representing an impact function; representing a horizontal sampling interval; Representing a vertical sampling interval; m and n each represent a summation count number.
6. The method for evaluating and promoting image cognitive interpretation information density according to claim 1 or 5, wherein S3 the processing algorithm is an adaptive priori regularization method, and the optimization model is as follows: (8) In formula (8): representing an original high interpretation quality image; p represents scene detail fidelity priori fitting parameters as optimization variables; Representing a point spread function; representing a minimized optimization; representing a regularized parameter matrix; Representing an imaging system degradation image; representing a smoothed 0-norm.

Description

Image cognition interpretation information density evaluation and promotion method Technical Field The invention relates to an image cognition interpretation information density evaluation and promotion method, and belongs to the technical field of optical imaging application. Background In the field of aviation and aerospace, various optical remote sensing images become an indispensable technical means in application fields such as natural monitoring, geographic exploration, environmental monitoring and the like by virtue of the capability of quickly and accurately acquiring ground object information. However, the optical remote sensing imaging process is restricted by multiple inherent factors such as atmospheric interference, system aberration, sensor noise, geometric radiation distortion and the like, so that the problems of fog blurring, detail degradation, noise artifact and the like of the image exist, and the effective extraction and application depth of ground feature information are severely limited. With the rapid increase of the requirements of high resolution and high timeliness on earth observation, unprecedented high standards are put forward on the definition, accuracy, fidelity and reliability of image data in the fields of homeland fine monitoring, disaster emergency response, environment dynamic evaluation, military target identification and the like. Currently, the links of an optical remote sensing imaging link are complex, the influence elements are highly coupled, the prior art lacks an effective information content evaluation method and a promotion means aiming at the loop interpretation application of people, so that the problems of noise amplification, artifact breeding and the like of a processed image are easy to occur, and the promotion and the guarantee of the quality of an original image become core technical bottlenecks for restricting accurate interpretation. The improvement of the cognitive interpretation quality of the optical remote sensing image is a key premise for releasing the value of the remote sensing big data, and the implementation effect of important strategic tasks such as global change research, resource management, national security and the like is directly influenced. In the prior art, although a certain progress is made in the fields of image restoration, super-resolution reconstruction and the like, a complete technical system from information density quantitative evaluation to targeted improvement of interpretation quality is not formed, and urgent requirements for improving the application efficiency of an optical satellite system are difficult to meet. The high-interpretation quality image can remarkably improve the classification precision of ground objects, the target recognition rate and the change detection sensitivity, and is a core support for promoting remote sensing application to change from qualitative to quantitative, from static to dynamic and from artificial to intelligent. Breaks through the effective information density evaluation of the optical remote sensing image and improves the technical bottleneck, not only relates to the availability and the credibility of remote sensing information products, but also improves the autonomous guarantee capability of the national space information and the strategic demands of preempting the high points of the remote sensing technology. Disclosure of Invention In order to solve the problems in the background technology, the invention provides an image cognition interpretation information density evaluation and promotion method. The invention realizes the aim by adopting the following technical scheme that the image cognition interpretation information density evaluation and promotion method comprises the following steps: S1, extracting an image interpretation quality influence element and a key parameter from a human eye cognition mechanism and an interpretation flow, and combing the association relation of the image interpretation quality influence element and the key parameter; s2, comprehensively considering key parameters and establishing a cognitive interpretation information density evaluation model; And S3, based on the cognition interpretation information density evaluation model, aiming at considering transfer function improvement, texture detail maintenance and noise artifact suppression, designing a processing algorithm to realize fine improvement of image interpretation quality. Further, the image interpretation quality influence element in S1 includes a shape, texture, scale, shadow, and activity feature of the object. Further, the key parameters of S1 include ground resolution, dynamic range, transfer function, ringing aliasing, and signal-to-noise ratio. Further, the transfer function is an optical remote sensing satellite on-orbit full-link normalized transfer function H: (1) In the formula (1): Representing the transfer function contribution of the optical system; representing the tra