Construction waste classification image recognition system

Abstract

The invention discloses a construction waste classification image recognition system, which comprises an image acquisition module and an image processing module, and is characterized in that the image acquisition module is used for acquiring to-be-recognized construction waste video images; the image processing module is connected with the image acquisition module, and the image processing module is used for extracting multiple groups of key frame images from the construction waste video images, comparing to-be-compared object information on each group of key frame images with recorded information in the construction waste object feature library, and if the to-be-compared object information on each group of key frame images is consistent with the recorded information in the construction waste object feature library, outputting the to-be-compared object information. And determining that the type of the to-be-compared object is the object type corresponding to the recorded information. The construction waste can be quickly identified and classified, and the cost can be saved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental protection, and in particular to a construction waste classification image recognition system. Background Art

[0002] Construction waste includes abandoned soil, materials and other waste generated during the construction, expansion and demolition of various buildings, pipe networks and decoration and renovation, and is mainly composed of inorganic impurities such as concrete, steel bars, slag, bricks, wood and plastic. Sorting is a key step in realizing the resource reuse of construction waste, and identification is an important part of realizing intelligent sorting.

[0003] The current identification technology used for construction waste is based on optical images such as visible light, multi-spectrum, infrared, and X-rays, supplemented by adding darkrooms and converting images into pseudo-color and grayscale images to improve imaging quality, and identification is performed through spectrally sensitive band matching, feature pixel matching, and pre-trained AI models.

[0004] With regard to the rapid identification of construction waste sorting, the existing technology has the following problems: first, it requires multi-spectral and X-ray related devices, which restricts the application cost at the construction site; second, it requires devices such as dark boxes, which affects the versatility of sorting and identification. Summary of the invention

[0005] The present invention aims to solve one of the technical problems in the related art to at least a certain extent. To this end, one object of the present invention is to provide a construction waste classification image recognition system suitable for rapid identification of construction waste generated during construction or demolition and renovation.

[0006] To achieve the above object, the present invention is implemented through the following technical solutions:

[0007] A construction waste classification image recognition system includes: an image acquisition module 10, used to acquire construction waste video images to be identified; an image processing module 20, connected to the image acquisition module 10, the image processing module 20 is used to extract multiple groups of key frame images from the construction waste video images, and compare the object information to be compared on each group of the key frame images with the recorded information in the construction waste object feature library, and when the comparison is consistent, determine that the type of the object to be compared is the object type corresponding to the recorded information.

[0008] Optionally, the image processing module 20 includes: a key frame extraction unit 21, connected to the image acquisition module 10, the key frame extraction unit 21 is used to extract multiple groups of key frame images from the construction waste video image, and form a first list; a classification and recognition unit 22, connected to the key frame extraction unit 21, the classification and recognition unit 22 is used to color classify the objects on each group of key frame images in the first list, and form a second list; a background elimination unit 23, connected to the classification and recognition unit 22, the background elimination unit 23 is used to eliminate the background objects on each group of key frame images in the second list, and form a third list; an object type determination unit 24, connected to the background elimination unit 23, the object type determination unit 24 is used to compare the information of the object to be compared on each group of key frame images in the third list with the recorded information in the construction waste object feature library, and when the comparison is consistent, determine that the type of the object to be compared is the object type corresponding to the recorded information.

[0009] Optionally, the key frame extraction unit 21 includes: a GPS data acquisition subunit 211, used to obtain the spatial historical path when the image acquisition module 10 acquires the construction waste video image; the key frame extraction unit 21 is specifically used to: extract multiple groups of key frame images from the construction waste video image by means of fixed-distance equal-division frames and constant-number equal-division frames; wherein, the fixed-distance equal-division frame method is to perform spatiotemporal equivalence on all frame images of the construction waste video image according to the spatial historical path, convert them to the time axis, and obtain corresponding multiple groups of key frame images; the constant-number equal-division frame method is to calculate the time interval constant based on the total number of images acquired from the construction waste video image and the total duration of the construction waste video image, and extract multiple groups of key frame images based on the time interval constant.

[0010] Optionally, the classification and identification unit 22 includes: a dot matrix extraction subunit 221, connected to the key frame extraction unit 21, and used to extract multiple groups of simplified dot matrices that are evenly distributed from each group of key frame images in the first list according to a preset accuracy; a data processing subunit 222, connected to the dot matrix extraction subunit 221, and used to calculate the RGB three-channel cubic difference of each adjacent point in the same row on each group of simplified dot matrices row by row, and when the RGB three-channel cubic difference is less than a first preset value, determine that the corresponding adjacent points belong to the same object, and input the dot matrix coordinates belonging to the same object into the object color list Lh.

[0011] Optionally, the data processing sub-unit 222 is also used to: calculate the RGB three-channel cubic difference of each adjacent point in the same column on each group of simplified dot matrices column by column, and when the RGB three-channel cubic difference is less than a first preset value, determine that the corresponding adjacent points belong to the same object, and input the dot matrix coordinates belonging to the same object into the object color list Lv; the data processing sub-unit 222 obtains the object recognition result of the simplified dot matrix based on the object recognition results in the object color list Lh and the object color list Lv, and forms the second list based on the recognition results of each group of simplified dot matrices on each group of key frame images.

[0012] Optionally, for each group of simplified dot matrices of each group of key frame images in the second list, the background elimination unit 23 is specifically used to: sequentially calculate the horizontal and vertical color coverage of the same object in each group of simplified dot matrices, and when the horizontal and vertical color coverage ratios calculated twice are both greater than a second preset value, determine that the object is an object to be eliminated, and eliminate the object to be eliminated from the corresponding key frame image.

[0013] Optionally, the object type determination unit 24 is specifically used to perform a point-by-point cubic difference comparison on the simplified dot matrix mean RGB information of the object to be compared in the third list with the RGB information in the construction waste object feature library, and when the number of points that pass the comparison is greater than the number of points that fail the comparison, and the ratio of the number of points that pass the comparison to the number of points that fail the comparison is greater than a third preset value, determine that the type of the object to be compared is the object type corresponding to the RGB information in the construction waste object feature library.

[0014] Optionally, the construction waste video image is a rectangular video image.

[0015] Optionally, the image acquisition module 10 includes but is not limited to: a mobile phone camera, a drone-mounted camera, and a digital camera.

[0016] Optionally, the image processing module 20 includes but is not limited to: a mobile terminal image processing software module and a computer terminal image processing software module.

[0017] The present invention has at least the following technical effects:

[0018] The present invention collects the video images of construction waste to be identified through the image acquisition module, and extracts multiple groups of key frame images from the construction waste video images through the image processing module, and then compares the color information of the objects to be compared on each group of key frame images with the color record information in the construction waste object feature library, so that the identification and classification of the objects can be completed quickly, and the image processing can be performed through the mobile terminal image processing software module, which effectively saves the processing cost.

[0019] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A structural block diagram of a construction waste classification image recognition system provided by an embodiment of the present invention;

[0021] Figure 2 A structural block diagram of a construction waste classification image recognition system provided by another embodiment of the present invention;

[0022] Figure 3 This is a structural block diagram of a construction waste classification image recognition system provided by another embodiment of the present invention. DETAILED DESCRIPTION

[0023] The present embodiment is described in detail below, and examples of the embodiment are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0024] The construction waste classification image recognition system of this embodiment is described below with reference to the accompanying drawings.

[0025] Figure 1 This is a structural block diagram of a construction waste classification image recognition system provided by an embodiment of the present invention. Figure 1 As shown, the system includes: an image acquisition module 10 and an image processing module 20.

[0026] Among them, the image acquisition module 10 is used to collect video images of construction waste to be identified; the image processing module 20 is connected to the image acquisition module 10, and the image processing module 20 is used to extract multiple groups of key frame images from the construction waste video images, and compare the object information to be compared on each group of key frame images with the recorded information in the construction waste object feature library, and when the comparison is consistent, determine that the type of the object to be compared is the object type corresponding to the recorded information.

[0027] The image acquisition module 10 includes but is not limited to: a mobile phone camera, a drone-mounted camera, and a digital camera. The image processing module 20 includes but is not limited to: a mobile terminal image processing software module and a computer terminal image processing software module. For example, a mobile phone APP (Application) and a PC (Personal Computer) image processing software client.

[0028] Specifically, the video image of the construction waste to be identified can be obtained by shooting through the image acquisition module 10, such as a mobile phone camera, a digital camera, an unmanned aerial vehicle camera, and a surveillance video probe, and the image is imported into the image processing module 20 (such as a mobile phone APP), wherein the construction waste video image is a video image of a rectangular frame. After the image processing module 20 obtains the construction waste video image, it extracts multiple groups of key frame images from the construction waste video image, and then compares the color information of each object on each group of key frame images with the color information recorded in the construction waste object feature library. When the comparison is consistent, it can be determined that the object to be compared is an object recorded in the construction waste object feature library, and it can be determined that the type of the object to be compared is the type of the object recorded in the construction waste object feature library to be compared with it.

[0029] Figure 2 This is a structural block diagram of a construction waste classification image recognition system provided by another embodiment of the present invention. Figure 2 As shown, the image processing module 20 includes: a key frame extraction unit 21, a classification and recognition unit 22, a background elimination unit 23 and an object type determination unit 24.

[0030] Among them, the key frame extraction unit 21 is connected to the image acquisition module 10, and the key frame extraction unit 21 is used to extract multiple groups of key frame images from the construction waste video image and form a first list; the classification and recognition unit 22 is connected to the key frame extraction unit 21, and the classification and recognition unit 22 is used to color classify the objects on each group of key frame images in the first list and form a second list; the background elimination unit 23 is connected to the classification and recognition unit 22, and the background elimination unit 23 is used to eliminate the background objects on each group of key frame images in the second list and form a third list; the object type determination unit 24 is connected to the background elimination unit 23, and the object type determination unit 24 is used to compare the information of the object to be compared on each group of key frame images in the third list with the recorded information in the construction waste object feature library, and when the comparison is consistent, determine that the type of the object to be compared is the object type corresponding to the recorded information.

[0031] Specifically, the key frame extraction unit 21 can extract multiple groups of key frame images from the construction waste video image, and form a total set of key frame images, i.e., a first list, which is recorded as Lp. After obtaining the first list Lp, the classification and recognition unit 22 performs color classification on the objects on each group of key frame images in the first list Lp to form a second list L1, in which the objects on each group of key frame images after color classification are recorded. The background elimination unit 23 obtains the second list L1, and deletes the background objects on each group of key frame images in the second list L1, such as the sky, water surface, or floor, and then forms a third list L2 based on the processed groups of key frame images. The object type determination unit 24 compares the RGB (Red Green Blue) information of the objects on each group of key frame images after color classification and background elimination in the third list L2 with the RGB information in the construction waste object feature library, and determines the type of the object to be compared when the comparison is consistent.

[0032] Figure 3 This is a structural block diagram of a construction waste classification image recognition system provided by another embodiment of the present invention. Figure 3 As shown, the key frame extraction unit 21 includes: a GPS positioning subunit 211.

[0033] In this embodiment, the GPS data acquisition subunit 211 is used to obtain the spatial historical path when the image acquisition module 10 collects the construction waste video image. The key frame extraction unit 21 is specifically used to extract multiple groups of key frame images from the construction waste video image by means of fixed-distance equal-division frames and fixed-number equal-division frames; wherein, the fixed-distance equal-division frame method is to perform spatiotemporal equivalence on all frame images of the construction waste video image according to the spatial historical path, and convert them to the time axis to obtain corresponding multiple groups of key frame images. The fixed-number equal-division frame method is to calculate the time interval constant based on the total number of images obtained from the construction waste video image and the total duration of the construction waste video image, and extract multiple groups of key frame images based on the time interval constant.

[0034] Specifically, the key frame images can be extracted by selecting one of the two methods: fixed-distance equally divided frames and fixed-number equally divided frames. It should be noted that when the video images of construction waste to be identified are captured by the image acquisition module 10, such as a mobile phone camera, a digital camera, an unmanned aerial vehicle camera, and a surveillance video probe, the spatial historical path when the video image is captured can be recorded by the GPS (Global Positioning System) program on the mobile phone (unmanned aerial vehicle). Further, before the key frame images are extracted by using fixed-distance equally divided frames, the GPS data acquisition subunit 211 can be used to collect and obtain the spatial historical path recorded in the GPS program, and convert it to the corresponding time axis according to the time and space equivalence to extract multiple groups of key frame images. The fixed-number equally divided frames are the total number of photos set according to the construction waste video images, and the time interval constant is calculated according to the total number of photos and the total length of the video, and then multiple groups of key frame images are extracted from the construction waste video images according to the time interval constant.

[0035] Please continue to refer to Figure 3 , the classification and identification unit 22 includes: a dot matrix extraction subunit 221 and a data processing subunit 222. In this embodiment, the dot matrix extraction subunit 221 is connected to the key frame extraction unit 21, and the dot matrix extraction subunit 221 is used to extract multiple groups of uniformly distributed simplified dot matrices from each group of key frame images in the first list Lp according to a preset accuracy; the data processing subunit 222 is connected to the dot matrix extraction subunit 221, and the data processing subunit 222 is used to calculate the RGB three-channel cubic difference of each adjacent point in the same row on each group of simplified dot matrices row by row, and when the RGB three-channel cubic difference is less than a first preset value, it is determined that the corresponding adjacent points belong to the same object, and the dot matrix coordinates belonging to the same object are input into the object color list Lh. The data processing subunit 222 is also used to: calculate the RGB three-channel cubic difference of each adjacent point in the same column on each group of simplified dot matrices column by column, and when the RGB three-channel cubic difference is less than a first preset value, determine that the corresponding adjacent points belong to the same object, and input the dot matrix coordinates belonging to the same object into the object color list Lv; the data processing subunit 222 obtains the object recognition result of the simplified dot matrix according to the object recognition results in the object color list Lh and the object color list Lv, and forms a second list L1 according to the recognition results of each group of simplified dot matrices on each group of key frame images.

[0036] Specifically, for each group of key frame images, the dot matrix extraction subunit 221 can extract multiple groups of uniformly distributed simplified dot matrices from all pixels of the group of key frame images (key frame images) according to a preset accuracy. For each group of simplified dot matrices, the RGB three-channel cubic difference between adjacent points in the same row on the dot matrix can be calculated row by row. If the RGB three-channel cubic difference is less than the first preset value, the judgment is passed, which means that the color of the pair of adjacent points is similar and the pair of adjacent points belongs to the same object. Otherwise, it is judged as different objects, and then the dot matrix coordinates belonging to the same object are input into the object color list Lh. Similarly, the RGB three-channel cubic difference of each adjacent point in the same column on the group of simplified dot matrices is calculated column by column, and a corresponding judgment is made, and then the dot matrix coordinates belonging to the same object are input into the object color list Lv. The data processing subunit 222 superimposes the calculation results in the object color list Lh and the object color list Lv to obtain the object recognition result of the group of simplified dot matrices in the key frame image, for example, multiple objects of different colors are recognized in the simplified dot matrix. Thus, the second list L1 can be formed according to the object recognition results of the multiple groups of simplified dot matrices in the multiple groups of key frame images. For example, the second list L1 records the object recognition results of the multiple groups of simplified dot matrices of the first group of key frame images, the object recognition results of the multiple groups of simplified dot matrices of the second group of key frame images, and so on.

[0037] In this embodiment, for each group of simplified dot matrices of each group of key frame images in the second list L1, the background elimination unit 23 is specifically used to: sequentially calculate the horizontal and vertical color coverage of the same object in each group of simplified dot matrices, and when the horizontal and vertical color coverage ratios calculated twice are both greater than a second preset value, determine that the object is an object to be eliminated, and eliminate the object to be eliminated from the corresponding key frame image.

[0038] Specifically, for each group of simplified dot matrices of each group of key frame images in the second list L1, the horizontal and vertical color coverage (such as dot ratio) of the same object in each group of simplified dot matrices is calculated from the outer layer to the inner layer. For example, the dot ratio of an object on the same horizontal / vertical column of the first group of simplified dot matrices is first calculated. If the dot ratio is greater than the second preset value, the judgment is passed. Then, the dot ratio of the object on the same horizontal / vertical column of the second group of simplified dot matrices is further calculated. If the dot ratio is greater than the second preset value, the object can be determined to be a background object, such as a base surface such as the sky, water surface or floor, and the pixel group corresponding to the dot matrix of the object is eliminated from the key frame image. The background elimination of all key frame images is completed in sequence, and the third list L2 is obtained.

[0039] In this embodiment, the object type determination unit 24 is specifically used to perform a point-by-point cubic difference comparison on the simplified dot matrix mean RGB information of the object to be compared in the third list L2 with the RGB information in the construction waste object feature library, and when the number of points that pass the comparison is greater than the number of points that fail the comparison, and the ratio of the number of points that pass the comparison to the number of points that fail the comparison is greater than a third preset value, determine that the type of the object to be compared is the type corresponding to the RGB information in the construction waste object feature library.

[0040] Specifically, the RGB information of the objects in each group of key frame images in the third list L2 can be compared with the RGB information in the construction waste object feature library by point cubic difference, and the type of the object to be compared can be determined. In this way, the type determination of all objects in the third list L2 can be completed.

[0041] In summary, the present invention collects the video images of construction waste to be identified through the image acquisition module, and extracts multiple groups of key frame images from the construction waste video images through the image processing module, and then compares the color information of the objects to be compared on each group of key frame images with the color record information in the construction waste object feature library, thereby quickly completing the identification and classification of the objects, and can perform image processing through the mobile terminal image processing software module, effectively saving processing costs.

[0042] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0043] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A construction waste classification image recognition system, characterized in that: include: An image acquisition module (10) is used to acquire video images of construction waste to be identified; An image processing module (20) is connected to the image acquisition module (10), and is used to extract multiple groups of key frame images from the construction waste video image, and compare the information of the object to be compared on each group of the key frame images with the recorded information in the construction waste object feature library, and when the comparison is consistent, determine that the type of the object to be compared is the object type corresponding to the recorded information.

2. The construction waste classification image recognition system according to claim 1, characterized in that: The image processing module (20) comprises: A key frame extraction unit (21) is connected to the image acquisition module (10), and the key frame extraction unit (21) is used to extract multiple groups of key frame images from the construction waste video image and form a first list; A classification and recognition unit (22) connected to the key frame extraction unit (21), the classification and recognition unit (22) being used to classify the colors of the objects on each group of key frame images in the first list and form a second list; A background elimination unit (23) connected to the classification and identification unit (22), the background elimination unit (23) being used to eliminate background objects on each group of key frame images in the second list and form a third list; An object type determination unit (24) is connected to the background elimination unit (23), and is used to compare the information of the object to be compared on each group of key frame images in the third list with the recorded information in the construction waste object feature library, and when the comparison is consistent, determine that the type of the object to be compared is the object type corresponding to the recorded information.

3. The construction waste classification image recognition system according to claim 2, characterized in that: The key frame extraction unit (21) comprises: A GPS data acquisition subunit (211) is used to obtain the spatial history path when the image acquisition module (10) acquires the construction waste video image; The key frame extraction unit (21) is specifically used to extract multiple groups of key frame images from the construction waste video image by means of fixed-distance equal-division frames and constant-number equal-division frames; wherein the fixed-distance equal-division frame method is to perform spatiotemporal equivalence on all frame images of the construction waste video image according to the spatial historical path, convert them to the time axis, and obtain corresponding multiple groups of key frame images; the constant-number equal-division frame method is to calculate a time interval constant based on the total number of images obtained from the construction waste video image and the total duration of the construction waste video image, and extract multiple groups of key frame images based on the time interval constant.

4. The construction waste classification image recognition system according to claim 2, characterized in that: The classification and identification unit (22) comprises: A dot matrix extraction subunit (221), connected to the key frame extraction unit (21), the dot matrix extraction subunit (221) being used to extract a plurality of groups of simplified dot matrices that are evenly distributed from each group of key frame images in the first list according to a preset accuracy; A data processing subunit (222) is connected to the dot matrix extraction subunit (221), and is used to calculate the RGB three-channel cubic difference of each adjacent point in the same row on each group of simplified dot matrix row by row, and when the RGB three-channel cubic difference is less than a first preset value, determine that the corresponding adjacent points belong to the same object, and input the dot matrix coordinates belonging to the same object into the object color list Lh.

5. The construction waste classification image recognition system according to claim 4, characterized in that: The data processing subunit (222) is also used for: The RGB three-channel cubic difference of each adjacent point in the same column on each group of simplified dot matrix is ​​calculated column by column, and when the RGB three-channel cubic difference is less than a first preset value, it is determined that the corresponding adjacent points belong to the same object, and the dot matrix coordinates belonging to the same object are input into the object color list Lv; the data processing subunit (222) obtains the object recognition result of the simplified dot matrix according to the object recognition results in the object color list Lh and the object color list Lv, and forms the second list according to the recognition results of each group of simplified dot matrix on each group of key frame images.

6. The construction waste classification image recognition system according to claim 2, characterized in that: For each group of simplified dot matrices of each group of key frame images in the second list, the background removal unit (23) is specifically used to: The horizontal and vertical color coverage of the same object in each group of simplified dot matrix is ​​calculated in turn, and when the horizontal and vertical color coverage obtained by two calculations are both greater than a second preset value, the object is determined to be an object to be eliminated, and the object to be eliminated is eliminated from the corresponding key frame image.

7. The construction waste classification image recognition system according to claim 2, characterized in that: The object type determination unit (24) is specifically used for: The simplified dot matrix mean RGB information of the object to be compared in the third list is compared with the RGB information in the construction waste object feature library by cubic difference of each point, and when the number of points that pass the comparison is greater than the number of points that fail the comparison, and the ratio of the number of points that pass the comparison to the number of points that fail the comparison is greater than a third preset value, it is determined that the type of the object to be compared is the object type corresponding to the RGB information in the construction waste object feature library.

8. The construction waste classification image recognition system according to claim 1, characterized in that: The construction waste video image is a rectangular video image.

9. The construction waste classification image recognition system according to claim 1, characterized in that: The image acquisition module (10) includes, but is not limited to: a mobile phone camera, a drone-mounted camera, and a digital camera.

10. The construction waste classification image recognition system according to claim 1, characterized in that: The image processing module (20) includes but is not limited to: a mobile terminal image processing software module and a computer terminal image processing software module.

Images