CN-121660924-B - Monitoring image defogging enhancement method in high-dust environment of beneficiation site

CN121660924BCN 121660924 BCN121660924 BCN 121660924BCN-121660924-B

Abstract

The invention belongs to the technical field of image processing, and in particular relates to a defogging and enhancing method for a monitoring image in a high-dust environment of a mineral processing site, which comprises the steps of carrying out smooth denoising on an original image by utilizing weighted guide filtering and obtaining global atmosphere light by a quadtree subdivision method; the method comprises the steps of constructing a same-color heterogeneous discrimination index combining a spectrum similarity factor and a texture confidence factor, calculating a pixel-level dynamic defogging coefficient based on the discrimination index, acquiring self-adaptive transmissivity by combining a dark channel prior, and finally restoring an image by using an atmospheric scattering model and carrying out contrast-limiting self-adaptive histogram equalization treatment. The invention effectively solves the problem of misjudgment caused by the fact that ore and dust in a beneficiation site are similar in color, realizes powerful defogging of a dust area and detail reservation of the ore area, and improves the definition and contrast of a monitoring image.

Inventors

LI TIANEN
SUN CHAOPENG
WANG ZHENGZHENG
ZHENG WENJIA

Assignees

西安天仁矿业信息科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260209

Claims (9)

1. The monitoring image defogging enhancement method in a high-dust environment of a beneficiation site is characterized by comprising the following steps of: The method comprises the steps of obtaining an original image of a beneficiation site, carrying out smoothing denoising treatment on the original image by using weighted guide filtering to obtain a pure image, carrying out iterative screening on the pure image by using a quadtree subdivision method, and calculating a global atmosphere light vector; according to the similarity of the RGB color vector of the local pixel of the pure image and the global atmosphere light vector, calculating the spectrum similarity factor of any pixel point of the pure image, In the formula (I), in the formula (II), Representing a spectral similarity factor of an x-th pixel point of the pure image; A color vector representing an xth pixel of the clear image; representing a global atmospheric light vector; representing a dot product operation symbol; modulo length symbols representing vectors; The method comprises the steps of representing a first tiny positive value, calculating texture confidence factors of any pixel point of a pure image according to the discrete degree of local gray level of the pure image, constructing homochromatic heterogeneous discrimination indexes of any pixel point of the pure image by combining the spectrum similarity factors and the texture confidence factors, linearly weighting basic defogging reference values according to the homochromatic heterogeneous discrimination indexes to obtain dynamic defogging coefficients of any pixel point of the pure image, processing the pure image by utilizing the dynamic defogging coefficients by combining a dark channel priori principle, and calculating the self-adaptive transmittance of any pixel point of the pure image; and inverting the pure image by utilizing the self-adaptive transmissivity and the global atmospheric light vector based on the atmospheric scattering physical model to obtain a physical restoration image, and carrying out detail enhancement on the physical restoration image to obtain a final enhanced image.
2. The method for defogging and enhancing the monitoring image in the high-dust environment of the beneficiation site according to claim 1, wherein the iterative screening of the pure image by using the quadtree subdivision method is performed, and the calculation of the global atmosphere light vector comprises the following steps: The method comprises the steps of dividing a pure image into four rectangular sub-blocks, calculating the difference value of the brightness mean value and the standard deviation of each rectangular sub-block, recording the difference value as the brightness stability index of the corresponding rectangular sub-block, taking the rectangular sub-block with the maximum brightness stability index as a target area of the next iteration, iterating the target area until the pixel number of the selected maximum rectangular sub-block is smaller than the size threshold for stopping iteration, and recording the selected maximum rectangular sub-block of the last iteration as a final selected sub-block; And calculating the brightness average value of RGB three channels of all pixels in the final selected sub-block to obtain a global atmosphere light vector, wherein the size threshold value for stopping iteration is a preset value.
3. The method for defogging enhancement of a monitoring image in a high dust environment of a beneficiation site according to claim 1, wherein the texture confidence factor satisfies the expression: ; in the formula, A texture confidence factor representing the x-th pixel point of the clean image; Representing the neighborhood gray variance of the x pixel point of the clean image; Representing the neighborhood gray average value of the x pixel point of the clean image; Representing a second slightly positive value.
4. The method for defogging and enhancing a monitoring image in a high-dust environment of a beneficiation site according to claim 1, wherein the method for constructing the homotopic heterogeneous discrimination index of any pixel point of a pure image by combining the spectrum similarity factor and the texture confidence factor comprises the following steps: Presetting a spectrum similarity weight and a texture confidence weight; and (3) carrying out difference on the product of the spectral similarity weight and the spectral similarity factor of any pixel point of the pure image and the product of the texture confidence weight and the texture confidence factor of the pixel point of the pure image to obtain the homotopic heterogeneous discriminant index of the pixel point of the pure image.
5. The method for enhancing defogging of monitoring images in a high-dust environment of a beneficiation site according to claim 1, wherein the obtaining of the dynamic defogging coefficient comprises the following steps: and adding the product of the sensitivity adjustment coefficient and the same-color heterogeneous discrimination index of any pixel point of the pure image with the basic defogging reference value to obtain the same-color heterogeneous discrimination index of the pixel point.
6. The monitored image defogging enhancement method under a high dust environment of a beneficiation site according to claim 1, wherein the adaptive transmittance satisfies the expression: ; in the formula, Representing the adaptive transmissivity of the x pixel point of the pure image; representing the dynamic defogging coefficient of the x pixel point of the pure image; representing that the xth pixel point of the pure image is at The number of channels; representing global atmospheric light vectors Channel number; Representing a minimum function; representing the minimum value operation in three channels of R, G and B; representing a third slightly positive value.
7. The method for defogging and enhancing the monitoring image in the high-dust environment of the beneficiation site according to claim 1, wherein the obtaining of the physical recovery image comprises the following steps: calculating any channel value of any pixel point of the physical restored image; all channel values of all pixels of the physical restoration image together form the physical restoration image.
8. The method for defogging and enhancing the monitoring image in the high-dust environment of the beneficiation site according to claim 7, wherein the value of any channel of any pixel point of the physical recovery image satisfies the expression: ; in the formula, C-channel value of the x-th pixel point of the physical restored image is represented; representing that the xth pixel point of the pure image is at The number of channels; representing global atmospheric light vectors Channel number; Representing the adaptive transmissivity of the x pixel point of the pure image; Representing a lower transmittance threshold; Representing the maximum function.
9. The method for defogging and enhancing a monitoring image in a high-dust environment of a beneficiation site according to claim 1, wherein the step of carrying out detail enhancement on a physical restoration image to obtain a final enhanced image comprises the following steps: And processing the restored image by using a limited contrast self-adaptive histogram equalization algorithm to obtain a final enhanced image.

Description

Monitoring image defogging enhancement method in high-dust environment of beneficiation site Technical Field The invention relates to the technical field of image processing. More particularly, the invention relates to a defogging enhancement method for a monitoring image in a high-dust environment of a beneficiation site. Background In a modern ore dressing production flow, in order to grasp the production states of key links such as crushing, screening and transferring in real time, an industrial video monitoring system is usually deployed on an operation site so as to realize remote monitoring on equipment operation, ore conveying and personnel safety. However, the working condition environment of the ore dressing site is complex, and a large amount of associated dust exists in the operation process. These high concentration suspended particles and aerosols are scattered around the equipment and ore, resulting in severe fogging of the images acquired by the monitoring probe, particularly in greatly reduced image contrast, color distortion and blurred details. This not only prevents the central office personnel from clearly observing the working conditions on site, but also causes the follow-up intelligent analysis algorithm based on machine vision to fail due to the too low image quality. In order to solve the above visual impairment, the prior art generally adopts a general image defogging algorithm based on the theory of dark channel prior and the like. However, in the specific scene of the ore dressing site, the problem of co-color interference often exists, namely the color of the ore is similar to that of dust. When such images are processed using conventional defogging techniques, it is difficult for the algorithm to distinguish between ore entities having the same color and suspended dust, and it is easy to misjudge the ore surface of high brightness as a high concentration haze layer. The false judgment can cause the algorithm to perform false defogging on the ore area in the actual treatment, thereby causing abnormal phenomena of ore surface brightness reduction, large-area blackening or color supersaturation and the like, finally causing the loss of key texture information such as cracks and granularity on the ore surface, and causing the enhanced image to lose monitoring value on production details. Disclosure of Invention In order to solve the technical problem that the monitoring image enhancement effect is poor in the high-dust environment of the ore dressing site, the invention provides a defogging enhancement method for the monitoring image in the high-dust environment of the ore dressing site, which comprises the following steps: The method comprises the steps of obtaining an original image of a beneficiation site, carrying out smoothing denoising treatment on the original image by utilizing weighted guide filtering to obtain a pure image, carrying out iterative screening on the pure image by utilizing a quadtree subdivision method to calculate a global atmosphere light vector, calculating spectrum similarity factors of any pixel point of the pure image according to the similarity of RGB color vectors of local pixels of the pure image and the global atmosphere light vector, calculating texture confidence factors of any pixel point of the pure image according to the discrete degree of local gray of the pure image, constructing a homochromatic heterogeneous discrimination index of any pixel point of the pure image by combining the spectrum similarity factors and the texture confidence factors, carrying out linear weighting on a basic defogging reference value according to the homochromatic heterogeneous discrimination index to obtain a dynamic defogging coefficient of any pixel point of the pure image, carrying out processing on the pure image by utilizing the dynamic defogging coefficient according to a dark channel priori principle, calculating self-adaptive transmittance of any pixel point of the pure image, carrying out inversion on the pure image by utilizing the self-adaptive transmittance and the global atmosphere light vector based on an atmospheric scattering physical model to obtain a physical restoration image, and carrying out detail enhancement on the physical restoration image to obtain a final enhanced image. The method can distinguish the regional difference of suspended dust and solid ore by constructing the homochromatic heterogeneous discrimination index containing the spectrum similarity factor and the texture confidence factor, and generate the dynamic defogging coefficient according to the regional difference, so that the algorithm can adaptively reduce defogging strength of an ore region while penetrating through massive dust interference, thereby completely retaining the color and texture details of the ore surface, avoiding image blackening or color distortion caused by excessive defogging, and improving the monitoring visibility in severe industrial environments.