CN107220946A - A kind of real-time eliminating method of bad lumpiness image on rock transportation band - Google Patents

Abstract

The invention belongs to the technical field of image processing, and discloses a real-time elimination method of a bad blockiness image on a rock conveyor belt, comprising: acquiring an RGB blockiness image; calculating the grayscale corresponding to the reduced grayscale image after smoothing and filtering The average gray value and relative variance of the image; set the first normal average gray threshold to remove some bad rock blockiness images; set the first normal relative variance threshold to reject some bad rock blocky images; set the second normal The average gray threshold and the second normal relative variance threshold remove some bad rock blockiness images; get the corresponding gradient image, calculate the gradient average value and gradient relative variance; set the average gradient threshold and gradient relative variance threshold to remove some bad rocks Looseness image: Obtain the next piece of lumpiness image and repeat the above steps to quickly and accurately detect bad rock lumpiness images on the conveyor belt.

Description

A Real-time Elimination Method of Bad Lumpy Image on Rock Conveyor Belt

technical field

The invention belongs to the technical field of image processing, and in particular relates to a real-time elimination method for bad lumpiness images on a rock conveyor belt, which is suitable for online detection and analysis of moving rock lumpiness images on the conveyor belt.

Background technique

In quarrying and mining engineering, the measurement of the size distribution of rock lumpiness is very important. Stone is a mixture of natural rock blocks and blasted and mechanically broken rock blocks, which are mainly used for construction, roads, railways and dams. In order to judge the quality of the stone, it is necessary to estimate the size and shape parameters of the stone particles. The size distribution of stone is not only a data used to evaluate product quality, but also important information for adjusting crusher or blasting production, for example: in quarrying production, adjusting the gap of crusher and adjusting the hole diameter of drilling in mining engineering, etc. . Crusher is usually set to produce stone within a strictly specified relatively narrow size range, such as from 16mm to 30mm. Usually a major indicator of crusher operation is the average size. In the automatic crushing control system, the feedback signal from the real-time system, including the average stone size, shows the actual progress of the crushing process on the line. In practical applications, the broken particles from the crusher are transported on a conveyor belt, and a CCD (charge coupled device, charge-coupled device) camera is placed above it to shoot downwards, and then image processing, segmentation and analysis are used to analyze the acquired images. Particles in are measured.

The existing technology mainly uses image processing and analysis and computer vision technology to process and analyze complex rock mass images at high speed and high precision, laying a new application foundation for improving the automatic monitoring and control level of mining and beneficiation production lines . The ore lumpiness image is the most complex multi-object image, because the color, grain size, shape, roughness, three-dimensional structure and other characteristics of the ore lumpiness make image processing, analysis and description far more difficult than other granular object images, so In this regard, pattern recognition, image analysis and machine vision are of great significance. The problem is: the image changes too much and too frequently, resulting in errors in image segmentation, resulting in wrong measurement and analysis results. In order to overcome this problem, the processed images should be selective, and try to remove those low-quality images before image processing For example: images without rock massiness, whiteboards made of reflections from rainwater conveyor belts, blurred images made of motion jitter, etc.

Contents of the invention

In view of the above problems, the purpose of the present invention is to provide a real-time method for removing bad rock lumpiness images on the conveyor belt, which can quickly and accurately detect bad rock lumpiness images on the conveyor belt, so as to solve the problems of image quality assessment and elimination , so that it can serve the monitoring and control of the industrial assembly line in mines and quarries.

In order to achieve the above object, the present invention adopts the following technical solutions to achieve.

A real-time method for removing bad lumpiness images on a rock conveyor belt, said method comprising the steps of:

Step 1, obtaining a blockiness image of moving rocks on the conveyor belt, the blockiness image is an RGB image;

Step 2, converting the RGB image into a corresponding grayscale image, and reducing the grayscale image according to a preset scale factor to obtain a reduced grayscale image;

Step 3, smoothing and filtering the reduced grayscale image to obtain a smoothed and filtered grayscale image;

Step 4, calculating the average gray value and relative variance of the gray image after smoothing and filtering;

Step 5, setting the first normal average grayscale threshold, if the average grayscale value of the smooth-filtered grayscale image is less than or equal to the first normal average grayscale threshold, then the smooth-filtered grayscale image The piece size image of the moving rock on the corresponding conveyor belt is a bad rock piece size image, and the bad rock piece size image is removed;

If the average gray value of the smooth-filtered gray image is greater than the first normal average gray threshold, proceed to step 6;

Step 6, setting the first normal relative variance threshold, if the relative variance of the smooth-filtered grayscale image is less than or equal to the first normal relative variance threshold, then the conveyor belt corresponding to the smooth-filtered grayscale image The piece size image of the moving rock is a bad rock piece size image, and the bad rock piece size image is removed;

If the relative variance of the smooth-filtered grayscale image is greater than the first normal relative variance threshold, proceed to step 7;

Step 7, setting a second normal average grayscale threshold and a second normal relative variance threshold, if the average grayscale value of the smooth-filtered grayscale image is less than or equal to the second normal average grayscale threshold, and the If the relative variance of the smooth-filtered grayscale image is less than or equal to the second normal relative variance threshold, the blockiness image of the moving rock on the conveyor belt corresponding to the smooth-filtered grayscale image is bad rock blockiness image, remove the bad rock blockiness image; otherwise, proceed to step 8;

Step 8, obtaining a corresponding gradient image according to the grayscale image after smoothing and filtering, and calculating the average gradient value and the relative gradient variance of the gradient image;

Step 9, setting the average gradient threshold and the gradient relative variance threshold, if the average gradient value of the gradient image is less than or equal to the gradient threshold, and the gradient relative variance of the gradient image is less than or equal to the gradient relative variance threshold, Then the piece size image of the moving rock on the conveyor belt corresponding to the gradient image is a bad rock piece size image, and the bad rock piece size image is removed;

Step 10, acquire the next blockiness image of the moving rock on the conveyor belt, and repeat step 2 to step 9 in sequence, so as to remove the bad blockiness image of the moving rock on the conveyor belt in real time.

The purpose of the present invention is to evaluate the quality of dynamic images, and can quickly remove bad-quality rock blockiness images, so as to ensure that the images to be processed and segmented and analyzed in the next step have good quality. Without this process, poor-quality images will cause difficulties and analysis errors in subsequent image processing, so that real-time online accurate detection results cannot be guaranteed. In order to meet the needs of real-time processing, this method analyzes the image quality in several steps, avoiding repeated calculations between steps, and the analysis and calculation methods are as simple and effective as possible, avoiding the use of complex calculations and analysis . It is very suitable for the application of ore lumpiness production site, and it is also easy to expand to other similar online particle detection, such as: dynamic flotation air bubbles, woody debris on the conveyor belt, moving grain particles and fruit image quality analysis with detection.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is a schematic flowchart of a real-time removal method for bad blockiness images on a rock conveyor belt provided by an embodiment of the present invention.

detailed description

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

An embodiment of the present invention provides a method for real-time removal of bad lumpiness images on a rock conveyor belt, as shown in FIG. 1 , the method includes the following steps:

Step 1, acquiring a blockiness image of moving rocks on the conveyor belt, where the blockiness image is an RGB image.

Generally speaking, in order to obtain long-term real-time fragmentation images of moving rocks on the conveyor belt, an ordinary CCD camera needs to be installed above the conveyor belt to obtain fragmentation images of rocks moving on the conveyor belt. In order to avoid uneven external light And instability, and in order to prevent dust, rain and snow, etc., it is necessary to set up a light box with a closed top and side, and install a uniform light source (light) and a CCD camera in the box, in order to clearly capture the movement (generally the movement speed is 2- 3 m/s) block size image, the CCD camera should be able to set the image acquisition parameters, which are often referred to as the shutter and aperture, and the acquired image can be transmitted to the computer through the connected image board for subsequent image processing.

Step 2, converting the RGB image into a corresponding grayscale image, and reducing the grayscale image according to a preset scale factor to obtain a reduced grayscale image.

Step 2 specifically includes:

Since the color of the rock mass is not obvious, in order to reduce the amount of calculation, the RGB image is converted into a corresponding grayscale image:

f(x,y) ₌ (f(x,y) _R +f(x,y) _G +f(x,y) _B )/3

Among them, f(x, y) _R , f(x, y) _G , f(x, y) _B respectively represent the red, green and blue pixel values of the pixel located at (x, y) in the RGB image, f( x, y) _gray represents the gray value of the pixel located at (x, y) in the gray image; x∈(1,...,2×M), y∈(1,...,2×N ), 2×M is the total number of pixels in the width dimension of the grayscale image, and 2×N is the total number of pixels in the height dimension of the grayscale image;

In order to eliminate the noise of black spots and bright spots and further reduce the workload of subsequent processing, the grayscale image is reduced; the preset scale factor is 1/4, then the grayscale image is reduced according to the scale factor, every The average gray value of four adjacent pixels is taken as the gray value of the reduced gray image at the corresponding position, so that the total number of pixels of the reduced gray image in the width dimension is M, and the total number of pixels in the height dimension is The number of pixels is N.

Step 3, smoothing and filtering the reduced grayscale image to obtain a smoothed and filtered grayscale image.

If the reduced grayscale image is smoothed by a conventional image smoothing algorithm (such as the neighborhood average method or Gaussian smoothing method, etc.), the weak boundary may be smoothed away. Therefore, the fractional integral smoothing method is used to remove the The noise in the reduced grayscale image, but in order to keep the smoothness of the image and keep the boundary between blocks from being lost, the traditional filter template is improved:

The fractional order integral smoothing method is used to smooth and filter the reduced grayscale image, and the filter coefficient h of the 5×5 template used is:

Then the gray-scale value F(x ₁ , y ₁ ) of the smooth-filtered gray-scale image at (x ₁ , y ₁ ) is: F(x ₁ , y ₁ )=(f(x ₁ , y ₁ )* h)/8, and the gray value of pixels in the left two columns, right two columns, upper two rows and lower two rows of the smoothed gray image is the same as the gray value of the corresponding position of the reduced gray image ;

It should be noted that the gray value F(x ₁ , y ₁ ) of the smoothed gray image at (x ₁ , y ₁ ) is: F(x ₁ , y ₁ )=(f(x ₁ , y ₁ )*h)/8, the filtering operation represented by this formula refers to obtaining 25 pixels centered on the pixel at (x ₁ , y ₁ ) of the grayscale image after smoothing and filtering, so that the 25 pixels Convolute with the filter coefficients respectively, so as to obtain the gray value F(x ₁ , y ₁ ) at (x ₁ , y ₁ ).

Among them, f(x ₁ , y ₁ ) represents the gray value at (x ₁ , y ₁ ) of the reduced gray-scale image, and x ₁ ∈ (0, ..., M), y ₁ ∈ ( 0,...,N).

Step 4, calculating the average gray value and relative variance of the smoothed and filtered gray image.

Step 4 specifically includes:

Calculate the average gray value y and the relative variance S _phase of the smooth filtered gray image:

S _phase =(s/v)×100

Among them, F(x ₁ , y ₁ ) represents the gray value of the smooth-filtered gray-scale image at (x ₁ , y ₁ ), x ₁ ∈ (0,..., M), y ₁ ∈ (0 ,..., N), the total number of pixels in the width dimension of the smoothed and filtered grayscale image is M, the total number of pixels in the height dimension is N, and S represents the variance of the smoothed and filtered grayscale image.

Step 5, setting the first normal average grayscale threshold, if the average grayscale value of the smooth-filtered grayscale image is less than or equal to the first normal average grayscale threshold, then the smooth-filtered grayscale image The piece size image of the moving rock on the corresponding conveyor belt is a bad rock piece size image, and the bad rock piece size image is removed;

If the average gray value of the smooth-filtered gray image is greater than the first normal average gray threshold, proceed to step 6.

Generally speaking, due to the influence of light and motion speed, the image quality of the blockiness of moving rocks will be low, and when the average gray value of the blockiness image is very low, the blockiness of rocks in the image will be blurred or dark gray or even black One piece (such as: non-blockiness image) cannot be followed by image processing and segmentation, so in order to avoid selecting such an image, it is necessary to set the first normal average gray threshold according to the site conditions (rock color, size, etc.).

Setting the first normal average grayscale threshold is specifically: artificially select a high-quality image of moving rocks, obtain the average grayscale value of the high-quality image, and set 30% of the average grayscale value of the high-quality image as the first normal average Gray threshold.

Step 6, setting the first normal relative variance threshold, if the relative variance of the smooth-filtered grayscale image is less than or equal to the first normal relative variance threshold, then the conveyor belt corresponding to the smooth-filtered grayscale image The piece size image of the moving rock is a bad rock piece size image, and the bad rock piece size image is removed;

If the relative variance of the smooth-filtered grayscale image is greater than the first normal relative variance threshold, proceed to step 7.

Although some bad-quality images can be eliminated in step 5, although the average gray value of the selected images is relatively high, there are still some low-quality images, such as: blurred images caused by motion speed, or blurred images caused by sudden Whiteboard images caused by strong light, etc., the variance of these images will be very low, so the variance of the image can be used to judge whether there is blockiness in the image: if the variance value is high, it proves that the gray level of the image has a large difference, that is, the blockiness is relatively large many.

The problem is: under different lighting conditions, there will be different mean values of image flat gray levels, and at this time, the same number and size of blocks will also lead to large differences in variance. In order to avoid such a problem that it is difficult to judge uniformly, the technical solution uses relative error to judge the image quality.

In step 6, setting the first normal relative variance threshold is specifically: artificially select a high-quality image of the moving rock, obtain the relative variance of the high-quality image, and set 40% of the relative variance of the high-quality image as the first normal relative variance threshold.

Step 7, setting a second normal average grayscale threshold and a second normal relative variance threshold, if the average grayscale value of the smooth-filtered grayscale image is less than or equal to the second normal average grayscale threshold, and the If the relative variance of the smooth-filtered grayscale image is less than or equal to the second normal relative variance threshold, the blockiness image of the moving rock on the conveyor belt corresponding to the smooth-filtered grayscale image is bad rock blockiness image, remove the bad rock blockiness image; otherwise, proceed to step 8.

Further, step 5 and step 6 both use a single index to judge the quality of the image, but in some cases it is necessary to combine two parameters to make the judgment reliable.

In step 7, setting the second normal average gray threshold and the second normal relative variance threshold are specifically:

Artificially select a high-quality image of moving rocks, obtain the average gray value and relative variance of the high-quality image, set 45% of the average gray value of the high-quality image as the second normal average gray threshold, and set the high-quality image 60% of the relative variance of is used as the second normal relative variance threshold.

Step 8: Obtain the corresponding gradient image according to the smoothed and filtered grayscale image, and calculate the gradient average value and the gradient relative variance of the gradient image.

It should be noted that through the elimination of the above steps 5, 6 and 7, about 70%-80% of poor quality images will be screened out, and the remaining 20%-30% of poor quality images can be passed through steps 8 and 7. Criteria in step 9 to eliminate.

Step 8 specifically includes:

Carry out first-order differentiation to the grayscale image after the smooth filter to obtain the corresponding gradient image; calculate the gradient mean value V _of the gradient image and the gradient relative variance S _phase :

S _{1 phase} = (s ₁ /v ₁ )×100

Among them, G(x ₂ , y ₂ ) represents the gradient value of the gradient image at (x ₂ , y ₂ ), x ₂ ∈ (0, ..., M), y ₂ ∈ (0, ..., N ), the total number of pixels of the gradient image in the width dimension is M, the total number of pixels in the height dimension is N, and S ₁ represents the variance of the gradient image.

Step 9, setting the average gradient threshold and the gradient relative variance threshold, if the average gradient value of the gradient image is less than or equal to the gradient threshold, and the gradient relative variance of the gradient image is less than or equal to the gradient relative variance threshold, Then the blockiness image of the moving rock on the conveyor belt corresponding to the gradient image is a bad rock blockiness image, and the bad rock blockiness image is removed.

In step 9, the average gradient threshold and the gradient relative variance threshold are set as follows:

Artificially select a high-quality image of moving rocks, obtain the average gradient threshold and gradient relative variance threshold of the gradient image corresponding to the high-quality image, set 50% of the average gradient threshold of the gradient image corresponding to the high-quality image as the average gradient threshold, set 60% of the gradient relative variance threshold of the gradient image corresponding to the high-quality image is determined as the gradient relative variance threshold.

Step 10, acquire the next blockiness image of the moving rock on the conveyor belt, and repeat step 2 to step 9 in sequence, so as to remove the bad blockiness image of the moving rock on the conveyor belt in real time.

The final remaining blocky image is an image with better quality. As long as the subsequent processing algorithm is suitable for the category of the image, there will be no large error in the subsequent processing, so that subsequent image segmentation and analysis can be performed.

Those of ordinary skill in the art can understand that all or part of the steps to realize the above method embodiments can be completed by hardware related to program instructions, and the aforementioned programs can be stored in computer-readable storage media. When the program is executed, the execution includes The steps of the above-mentioned method embodiments; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other various media that can store program codes.

The above is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Anyone skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present invention. Should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims.

Claims (9)

1. a real-time method for removing bad lumpiness images on a rock conveyor belt, characterized in that, the method comprises the steps: Step 1, obtaining a blockiness image of moving rocks on the conveyor belt, the blockiness image is an RGB image; Step 2, converting the RGB image into a corresponding grayscale image, and reducing the grayscale image according to a preset scale factor to obtain a reduced grayscale image; Step 3, smoothing and filtering the reduced grayscale image to obtain a smoothed and filtered grayscale image; Step 4, calculating the average gray value and relative variance of the gray image after smoothing and filtering; Step 5, setting the first normal average grayscale threshold, if the average grayscale value of the smooth-filtered grayscale image is less than or equal to the first normal average grayscale threshold, then the smooth-filtered grayscale image The piece size image of the moving rock on the corresponding conveyor belt is a bad rock piece size image, and the bad rock piece size image is removed; If the average gray value of the smooth-filtered gray image is greater than the first normal average gray threshold, proceed to step 6; Step 6, setting the first normal relative variance threshold, if the relative variance of the smooth-filtered grayscale image is less than or equal to the first normal relative variance threshold, then the conveyor belt corresponding to the smooth-filtered grayscale image The piece size image of the moving rock is a bad rock piece size image, and the bad rock piece size image is removed; If the relative variance of the smooth-filtered grayscale image is greater than the first normal relative variance threshold, proceed to step 7; Step 7, setting a second normal average grayscale threshold and a second normal relative variance threshold, if the average grayscale value of the smooth-filtered grayscale image is less than or equal to the second normal average grayscale threshold, and the If the relative variance of the smooth-filtered grayscale image is less than or equal to the second normal relative variance threshold, the blockiness image of the moving rock on the conveyor belt corresponding to the smooth-filtered grayscale image is bad rock blockiness image, remove the bad rock blockiness image; otherwise, proceed to step 8; Step 8, obtaining a corresponding gradient image according to the grayscale image after smoothing and filtering, and calculating the average gradient value and the relative gradient variance of the gradient image; Step 9, setting the average gradient threshold and the gradient relative variance threshold, if the average gradient value of the gradient image is less than or equal to the gradient threshold, and the gradient relative variance of the gradient image is less than or equal to the gradient relative variance threshold, Then the piece size image of the moving rock on the conveyor belt corresponding to the gradient image is a bad rock piece size image, and the bad rock piece size image is removed; Step 10, acquire the next blockiness image of the moving rock on the conveyor belt, and repeat step 2 to step 9 in sequence, so as to remove the bad blockiness image of the moving rock on the conveyor belt in real time. 2. the real-time removal method of bad lumpiness image on a kind of rock conveyor belt according to claim 1, it is characterized in that, step 2 specifically comprises: Convert said RGB image to the corresponding grayscale image: f(x,y) ₌ (f(x,y) _R +f(x,y) _G +f(x,y) _B )/3 Among them, f(x,y) _R , f(x,y) _G , f(x,y) _B represent the red, green, and blue pixel values of the pixel at (x,y) in the RGB image respectively, and f( x, y) _gray represents the gray value of the pixel located at (x, y) in the gray image; x∈(1,...,2×M), y∈(1,...,2×N ), 2×M is the total number of pixels in the width dimension of the grayscale image, and 2×N is the total number of pixels in the height dimension of the grayscale image; If the preset scale factor is 1/4, the grayscale image is reduced according to the scale factor, and the grayscale average value of every four adjacent pixels is taken as the value of the reduced grayscale image at the corresponding position. Gray value, so that the total number of pixels in the width dimension of the reduced gray-scale image is M, and the total number of pixels in the height dimension is N. 3. the real-time removal method of bad lumpiness image on a kind of rock conveyor belt according to claim 1, it is characterized in that, step 3 specifically comprises: The fractional order integral smoothing method is used to smooth and filter the reduced grayscale image, and the filter coefficient of the 5×5 template used is: <mrow> <mi>h</mi> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mn>0.09375</mn> </mtd> <mtd> <mn>0.125</mn> </mtd> <mtd> <mn>0.09375</mn> </mtd> <mtd> <mn>0</mn> </mtd> </mtr> <mtr> <mtd> <mn>0.09375</mn> </mtd> <mtd> <mn>0.3125</mn> </mtd> <mtd> <mn>0.5</mn> </mtd> <mtd> <mn>0.3125</mn> </mtd> <mtd> <mn>0.09375</mn> </mtd> </mtr> <mtr> <mtd> <mn>0.125</mn> </mtd> <mtd> <mn>0.5</mn> </mtd> <mtd> <mn>2</mn> </mtd> <mtd> <mn>0.5</mn> </mtd> <mtd> <mn>0.125</mn> </mtd> </mtr> <mtr> <mtd> <mn>0.09375</mn> </mtd> <mtd> <mn>0.3125</mn> </mtd> <mtd> <mn>0.5</mn> </mtd> <mtd> <mn>0.3125</mn> </mtd> <mtd> <mn>0.09375</mn> </mtd> </mtr> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mrow> <mo>-</mo> <mn>0.09375</mn> </mrow> </mtd> <mtd> <mrow> <mo>-</mo> <mn>0.125</mn> </mrow> </mtd> <mtd> <mrow> <mo>-</mo> <mn>0.09375</mn> </mrow> </mtd> <mtd> <mn>0</mn> </mtd> </mtr> </mtable> </mfenced> </mrow> Then the gray value F(x ₁ , y ₁ ) of the gray image after smoothing and filtering at (x ₁ , y ₁ ) is: F(x ₁ , y1)=(f(x ₁ ,y ₁ )*h )/8, and the grayscale values of the left two columns, right two columns, upper two rows and lower two rows of pixels of the smooth filtered grayscale image are the same as the grayscale values at the corresponding positions of the reduced grayscale image; Among them, f(x ₁ ,y ₁ ) represents the gray value at (x ₁ ,y ₁ ) of the reduced gray-scale image, and x ₁ ∈(0,...,M), y ₁ ∈( 0,...,N). 4. the real-time removal method of bad lumpiness image on a kind of rock conveyor belt according to claim 1, it is characterized in that, step 4 specifically comprises: Calculate the average gray value V and relative variance S _of the smooth-filtered gray-scale image: <mrow> <mi>V</mi> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>x</mi> <mn>1</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>M</mi> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>y</mi> <mn>1</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <mi>F</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>y</mi> <mn>1</mn> </msub> <mo>)</mo> </mrow> <mo>/</mo> <mi>M</mi> <mo>&amp;times;</mo> <mi>N</mi> </mrow> <mrow> <mi>S</mi> <mo>=</mo> <msqrt> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>x</mi> <mn>1</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>M</mi> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>y</mi> <mn>1</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msup> <mrow> <mo>(</mo> <mi>F</mi> <mo>(</mo> <mrow> <msub> <mi>x</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>y</mi> <mn>1</mn> </msub> </mrow> <mo>)</mo> <mo>-</mo> <mi>V</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>/</mo> <mrow> <mo>(</mo> <mi>M</mi> <mo>&amp;times;</mo> <mi>N</mi> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </msqrt> </mrow> S _phase ＝(S/ _V )×100 Among them, F(x ₁ ,y ₁ ) represents the gray value of the smoothed and filtered gray image at (x ₁ ,y ₁ ), x ₁ ∈(0,...,M), y ₁ ∈(0 ,...,N), the total number of pixels in the width dimension of the smoothed and filtered grayscale image is M, and the total number of pixels in the height dimension is N, and S represents the variance of the smoothed and filtered grayscale image. 5. The real-time method for removing bad blockiness images on a rock conveyor belt according to claim 1, characterized in that, in step 5, setting the first normal average gray threshold is specifically: artificially selecting one of the moving rocks A high-quality image is obtained, the average gray value of the high-quality image is obtained, and 30% of the average gray value of the high-quality image is set as the first normal average gray value threshold. 6. The real-time method for removing bad blockiness images on a rock conveyor belt according to claim 1, wherein in step 6, setting the first normal relative variance threshold is specifically: artificially selecting a picture of moving rocks For a high-quality image, the relative variance of the high-quality image is obtained, and 40% of the relative variance of the high-quality image is set as the first normal relative variance threshold. 7. The real-time method for removing bad blockiness images on a rock conveyor belt according to claim 1, wherein in step 7, setting the second normal average gray threshold and the second normal relative variance threshold is specifically : Artificially select a high-quality image of moving rocks, obtain the average gray value and relative variance of the high-quality image, set 45% of the average gray value of the high-quality image as the second normal average gray threshold, and set the high-quality image 60% of the relative variance of is used as the second normal relative variance threshold. 8. The real-time method for removing bad lumpiness images on a rock conveyor belt according to claim 1, wherein step 8 specifically comprises: Carry out first-order differentiation to the grayscale image after the smooth filter to obtain the corresponding gradient image; calculate the gradient mean value V _of the gradient image and the gradient relative variance S _phase : <mrow> <msub> <mi>V</mi> <mn>1</mn> </msub> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>x</mi> <mn>2</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>M</mi> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>y</mi> <mn>2</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <mi>G</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mn>2</mn> </msub> <mo>,</mo> <msub> <mi>y</mi> <mn>2</mn> </msub> <mo>)</mo> </mrow> <mo>/</mo> <mi>M</mi> <mo>&amp;times;</mo> <mi>N</mi> </mrow> <mrow> <msub> <mi>S</mi> <mn>1</mn> </msub> <mo>=</mo> <msqrt> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>x</mi> <mn>2</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>M</mi> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <msub> <mi>y</mi> <mn>2</mn> </msub> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msup> <mrow> <mo>(</mo> <mi>G</mi> <mo>(</mo> <mrow> <msub> <mi>x</mi> <mn>2</mn> </msub> <mo>,</mo> <msub> <mi>y</mi> <mn>2</mn> </msub> </mrow> <mo>)</mo> <mo>-</mo> <msub> <mi>V</mi> <mn>1</mn> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>/</mo> <mrow> <mo>(</mo> <mi>M</mi> <mo>&amp;times;</mo> <mi>N</mi> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </msqrt> </mrow> S _{1 phase} = (S ₁ /V ₁ )×100 Among them, G(x ₂ ,y ₂ ) represents the gradient value of the gradient image at (x ₂ ,y ₂ ), x ₂ ∈(0,...,M), y ₂ ∈(0,...,N ), the total number of pixels of the gradient image in the width dimension is M, the total number of pixels in the height dimension is N, and S ₁ represents the variance of the gradient image. 9. The real-time method for removing bad blockiness images on a rock conveyor belt according to claim 1, wherein in step 9, setting the average gradient threshold and the gradient relative variance threshold is specifically: Artificially select a high-quality image of moving rocks, obtain the average gradient threshold and gradient relative variance threshold of the gradient image corresponding to the high-quality image, set 50% of the average gradient threshold of the gradient image corresponding to the high-quality image as the average gradient threshold, set 60% of the gradient relative variance threshold of the gradient image corresponding to the high-quality image is determined as the gradient relative variance threshold.