Method for identifying manganese block region based on region growing algorithm of color features

Abstract

The invention discloses a method for identifying a manganese block region based on a region growing algorithm of color features, and the method comprises the following steps: S1, carrying out the preprocessing of a shot image: adjusting the pixels of a reflective region and a black-yellow region through a color constancy algorithm, and converting the image into a gray-scale image through the classification gray-scale transformation; s2, selecting a seed area for area growth; s3, performing morphological opening operation on the grown region to remove slender branches and fuse cracks, and performing region merging, and S4, positioning a manganese block region: marking the merged region, performing rectangular frame positioning according to extreme values, namely min (xi), max (xi), min (y) and max (y), of region pixel points during growth, and judging whether the area of a rectangular frame is greater than St1 or not; st1 represents the effective beating area of the mechanical hammer, if the effective beating area is larger than the effective beating area, the rectangular frames are segmented, and the centroid of each rectangular frame is found.

Description

Technical Field

[0001] The invention relates to the field of image recognition, and in particular to a method for identifying a manganese block region using a region growing algorithm based on color features. Background Art

[0002] The task of object detection is to find all the objects of interest in the image and determine their categories and locations. It is one of the core issues in the field of computer vision. Since various objects have different appearances, shapes and postures, and are interfered by factors such as illumination and occlusion during imaging, object detection has always been the most challenging problem in the field of computer vision. At present, the use of machine learning methods to identify targets has penetrated into production and life. The basic idea of ​​the region growing method is to collect pixels with similar properties to form a region with maximum consistency. According to the given consistent similarity criterion, based on the pre-selected initial seed point, the pixels in the neighborhood of the seed point that meet the similarity criterion are merged with the seed pixel, and the new growth process is repeated until no pixels that meet the criterion conditions can be included. The key to region growing is the selection of seed points, the consistent similarity criterion and the determination of termination conditions. The image segmentation method of region growing considers both the similarity of pixels and the adjacency of pixels in space, which can effectively eliminate the interference of isolated noise, especially the coherent speckle noise of alternating light and dark, and has strong robustness and high segmentation accuracy.

[0003] There are two main ways to extract manganese metal: thermal method (fire method) and electrolytic method (wet method). The purity of thermal method production (manganese metal) does not exceed 95-98%, while pure manganese metal is prepared by electrolysis (electrolytic manganese metal), and its purity can reach 99.7-99.9% or more. Now, electrolytic production has become the main way to produce manganese metal. Electrolytic manganese metal is obtained by acid leaching manganese ore to obtain manganese salt, and then sent to the electrolytic cell for electrolysis to precipitate the elemental metal. In production, electrolytic manganese is decomposed onto the background plate, and the position of the manganese block on the background plate (there will be multiple blocks on the plate) is determined by taking an image with a camera, and then the manganese block area is hit by a mechanical hammer to knock it down to obtain the manganese block.

[0004] To perform a fixed-point tapping of manganese blocks by means of a mechanical hammer, the position information needs to be transmitted to the mechanical hammer, and the calculation results need to be obtained as quickly as possible in the pipeline processing. However, the use of deep learning methods will reduce the calculation speed, affect the speed of the mechanical hammer, and thus affect the production of the manganese blocks. In addition, the background plate will be worn out after long-term use, and the manganese block may also leave residues on the background plate, which will cause the background plate to reflect light, and there will be black and yellow stains on the pixels in the manganese block area. Both reflections and stains are noise. If the noise is not eliminated, in the process of identifying the captured pictures, the pixels in the reflective area will be closer to the pixel values ​​in the manganese block area after grayscale processing, and the black and yellow area will be lower than the value range of the manganese block pixels, which is difficult to eliminate in the subsequent judgment conditions for regional growth, affecting the accuracy of manganese block area recognition. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to address the deficiencies in the prior art and provide a method for identifying manganese block regions using a region growing algorithm based on color features.

[0006] Technical solution: A method for identifying manganese block regions using a region growing algorithm based on color features, comprising the following steps:

[0007] S1. Preprocess the captured images:

[0008] Divide the image into N blocks and calculate the pixel mean Mp(i) of each block; t1 Determine the reflective area R(i), by the pixel threshold (P t3 ,P t4 ) determines the black-yellow area B(i); adjusts the pixels of the reflective area and the black-yellow area through the color constancy algorithm, and then converts the image into a grayscale image using the classified grayscale transformation; wherein, P t1 represents the maximum value of the pixel range of the manganese block area; R(i) represents the i-th reflective area; P t3 Indicates the minimum value of pixels in the black and yellow area; P t4 represents the maximum value of the pixels in the black-yellow area; B(i) represents the i-th black-yellow area;

[0009] S2. Select seed regions for region growing:

[0010] Select the area pixel mean Mp(i) in S1 that is greater than the threshold P t2 At the same time, it is less than P t1 The connected area Sed(i) is sorted in descending order, and the area with the largest pixel mean is selected as the seed area for region growth, that is, Mp(i)<P t1 &&Mp(i)>P t2 is the seed area, where growth begins; t2represents the minimum pixel range of the manganese block area, Sed(i) represents the i-th seed area, and Mp(i) represents the pixel mean of the i-th block area;

[0011] S3, perform morphological opening operation on the grown area to remove the thin and long branches and merge the cracks, and finally calculate the distance D ij and area S t1 Perform region merging, where D t1 represents the centroid distance threshold; S t1 Indicates the effective area struck by the mechanical hammer;

[0012] S4. Positioning the manganese block area:

[0013] Mark the merged area according to the extreme value of the pixel point in the area during growth, i.e., min(x i ),max(x i ),min(y i ),max(y i ) to locate the rectangular frame and determine whether the area of ​​the rectangular frame is greater than S t1 If it is greater than, the rectangular frame is divided and the centroids of each rectangular frame are found respectively; the centroids of each rectangular frame are the centroids of the target manganese block area (xi, yi); among them, min(x i ),max(x i ),min(y i ),max(y i ) represents the coordinate extreme value of the pixels identified as manganese blocks in a given area in the two-dimensional plane.

[0014] Furthermore, in S1, the color constancy algorithm is a classified weighted gray world algorithm CWGW.

[0015] Furthermore, the specific steps of S1 are as follows:

[0016] S101, calculating the average pixel value of the region:

[0017] Divide the image into N rectangular regions and calculate the pixel mean Mp(i) of the rectangular region N(i);

[0018] Where N(i) represents the i-th area, and Mp(i) represents the pixel mean of the i-th area;

[0019] S102, determining whether it is a reflective area or a dark-yellow area;

[0020] When the pixel mean Mp(i) is greater than the pixel range (Pt2, Pt1) of the manganese block area, that is, greater than the threshold P t1 , then N(i) is considered to be the reflective area R(i); when the pixel mean Mp(i) belongs to the black-yellow area pixel range (Pt3 ,P t4 ), then N(i) is considered to be the black-yellow area B(i); if Mp(i) of N(i) is less than the pixel threshold P t1 , then the area is considered to be a manganese block area or other area where the reflection is not serious;

[0021] Among them, R(i) represents the i-th reflective area, and the pixel range of the manganese block (P t2 ,P t1 ) is obtained from sample statistics, P t1 Indicates the maximum pixel range of the manganese block, P t2 Indicates the minimum pixel range of the manganese block area; the pixel range of the black and yellow area (P t3 ,P t4 ) is obtained from sample statistics, P t3 Represents the minimum value of pixels in the black and yellow area, P t4 represents the maximum value of the pixels in the black-yellow area; B(i) represents the i-th black-yellow area;

[0022] S103, using the classified weighted gray world algorithm CWGW to adjust the pixels in the reflective area and the black-yellow area;

[0023] S104, grayscale conversion of the image: grayscale the reflective area R(i) after the pixels are adjusted in S103 using the average value method, and grayscale the other areas using the weighted average method.

[0024] Further, S103, the method of using the classified weighted gray world algorithm (CWGW) to adjust the pixels in the reflective area and the dark yellow area is:

[0025] 1) Calculate R according to the following formula new , G new , B new :

[0026]

[0027] Among them, R new , G new , B new Respectively represent the adjusted pixel R value, G value, and B value; θ is a smaller number used to appropriately increase the values ​​of the R channel and the G channel; R r , G r , B r Respectively represent the R value, G value, and B value of the original image color in the area with severe reflection;

[0028] 2) R new , G new , B new The pixels are merged to get the pixels after removing the reflection and black-yellow.

[0029] Furthermore, in S104, the grayscale transformation of the image is performed using the following formula:

[0030]

[0031] Among them, x, y represent pixel coordinates; Gray(x,y) represents the grayscale value at the coordinates (x,y); R(x,y) represents the value of the R channel at the coordinates (x,y); G(x,y) represents the value of the R channel at the coordinates (x,y); B(x,y) represents the value of the R channel at the coordinates (x,y).

[0032] Furthermore, the S2 specifically comprises the following steps:

[0033] S201: In the result Mp(i) calculated in S101, the mean value of the pixels in the region (P t2 ,P t1 ) is used as the seed region of the region growing algorithm;

[0034] S202: In the process of region growing, the pixel is (P t2 ,P t1 ) range as the growth condition. At the same time, if the pixels of two regions intersect, the two regions are merged; that is,

[0035]

[0036] Where x, y represent the coordinates of the pixel, Mk(i) represents the known i-th manganese block area (initially the seed area), It means merging the i-th manganese block area into the j-th manganese block area, and SMk(i) represents the area of ​​the i-th known manganese block area.

[0037] Furthermore, the calculation formula in S3 is as follows:

[0038]

[0039]

[0040] The opening operation is to first perform the corrosion operation and then the expansion operation. S represents the image area to be processed; A represents the structural element; D ij represents the distance between the centroid of the i-th area and the j-th area, D t1 represents the centroid distance threshold; S t1 It indicates the effective area struck by the mechanical hammer, and Merge means the merge operation.

[0041] Beneficial effect: The method of the present invention has a faster calculation speed than the deep learning method. In the region growing algorithm, the color feature is used as the growth threshold, the maximum and minimum thresholds are adopted, and the seed area is used as the seed point for region growth, which reduces the time required for the region growing algorithm to determine the seed point. At the same time, it reduces the time required in the region growing process, can better identify the pixels in the manganese block area, and improve the accuracy of manganese block area identification.

[0042] The classification weighted gray world algorithm (CWGW) used in the present invention can perform different data processing according to different regions, improving the original gray world algorithm that processes all pixels. The gray world algorithm is not rich in image scene colors, especially when large monochromatic objects appear, the algorithm often fails; in the present invention, there are large areas of background color and manganese block areas. If the gray world algorithm is used to process them, it is very likely that a lot of calculations will be wasted without any effect. The classification weighted gray world algorithm (CWGW) used in the present invention only processes the reflective and black and yellow parts, reducing the time required for calculation and having higher accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a flow chart of the method of the present invention;

[0044] Figure 2 Flow chart of the preprocessing process (i.e., gray world algorithm and graying) in the method of the present invention;

[0045] Figure 3 Flow chart of the region growing algorithm in the method of the present invention;

[0046] Figure 4 This is an example of a manganese block area. DETAILED DESCRIPTION

[0047] The technical solution of the present invention is described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the embodiments.

[0048] like Figure 1 As shown, the present invention provides a method for identifying a manganese block region using a region growing algorithm based on color features, comprising the following steps:

[0049] S1. Preprocessing the captured image, including color constancy processing and grayscale conversion;

[0050] The image is divided into N blocks, and the pixel threshold P is used t1 Determine the reflective area R(i); adjust the pixels of the reflective area and the black-yellow area through the color constancy algorithm to reduce the influence of illumination and the special cases in the manganese block, and then use the classified grayscale transformation to convert the image into a grayscale image; where P t1It represents the maximum value of the pixel range of the manganese block area, which is obtained by measuring the sample data (because the background plate is black and the manganese block is grayish white, if there is a higher pixel value than the manganese block, it can be basically judged as a reflective area, that is, Pt1 here is the maximum pixel value of the manganese block area, and it is also the minimum pixel value of the reflective area.), S(i) represents the area of ​​the ith block area, and R(i) represents the ith reflective area.

[0051] The color constancy algorithm is the Classification Weighted Gray World Algorithm CWGW. Since the grayscale image has only one channel, the processing speed is faster than the three channels of the color image.

[0052] The specific steps in S1 are as follows:

[0053] S101, calculating the average pixel value of the region:

[0054] Divide the image into N rectangular regions and calculate the pixel mean Mp(i) of the rectangular region N(i);

[0055] Among them, N(i) represents the i-th block area, and Mp(i) represents the pixel mean of the i-th block area.

[0056] S102, determining whether it is a reflective area or a dark-yellow area;

[0057] When the pixel mean Mp(i) is greater than the pixel range of the manganese block area (P t2 ,P t1 ), which is greater than the threshold P t1 , then N(i) is considered to be the reflective area R(i); when the pixel mean Mp(i) belongs to the black-yellow area pixel range (P t3 ,P t4 ), then N(i) is considered to be the black-yellow area B(i); if Mp(i) of N(i) is less than the pixel threshold P t1 , then the area is considered to be a manganese block area or other area where the reflection is not serious;

[0058] Among them, R(i) represents the i-th reflective area, and the pixel range of the manganese block (P t2 ,P t1 ) is obtained from sample statistics, P t1 represents the maximum value of the manganese block pixel range (which is also the minimum value of the reflective area pixel), P t2 Indicates the minimum pixel range of the manganese block area and the pixel range of the black and yellow area (P t3 ,P t4 ) is obtained from sample statistics, P t3 Indicates the minimum value of pixels in the black and yellow area, P t4 It represents the maximum value of the pixels in the black and yellow area, and B(i) represents the i-th black and yellow area.

[0059] S103, using a classified weighted gray world algorithm (CWGW) to adjust the pixels in the reflective area and the black-yellow area;

[0060] Each pixel in the reflective area and the black-yellow area is used as the parameter input of the classification weighted gray world algorithm (CWGW) to reduce the influence of illumination and the special cases in the manganese block;

[0061] The Gray World Algorithm (GW) usually adjusts the entire image, while the Classification Weighted Gray World Algorithm (CWGW) used in the present invention processes the reflective area and the black and yellow stain area differently, avoiding the situation where the algorithm fails due to large monochromatic objects.

[0062] In the reflection area of ​​the classified weighted gray world algorithm, the RGB mean (R r ,G r ,B r ) as the new value; through the sample observation, the manganese block area has black and yellow stains that affect the recognition result, so in this process, the proportion of black and yellow is reduced and the proportion of gray and white is increased. Among them, R r , G r , B r They respectively represent the R value, G value, and B value of the original image color in the area with severe reflection.

[0063] The method of using the classified weighted gray world algorithm (CWGW) to adjust the pixels in the reflective area and the black-yellow area is as follows:

[0064] 1) Calculate R according to the following formula new , G new , B new :

[0065]

[0066] Among them, R new , G new , B new Respectively represent the adjusted pixel R value, G value, and B value. θ is a smaller number used to appropriately increase the values ​​of the R and G channels. r , G r , B r Respectively represent the R value, G value, and B value of the original image color in the area with severe reflection;

[0067] 2) R new , G new , B new The pixels are merged to get the pixels after removing the reflection and black-yellow.

[0068] S104: grayscale conversion of the image: grayscale the reflective area R(i) after adjusting the pixels in S103 using the average method, and grayscale the other areas using the weighted average method. That is:

[0069]

[0070] Among them, x, y represent pixel coordinates; Gray(x,y) represents the grayscale value at the coordinates (x,y); R(x,y) represents the value of the R channel at the coordinates (x,y); G(x,y) represents the value of the R channel at the coordinates (x,y); B(x,y) represents the value of the R channel at the coordinates (x,y).

[0071] S2, select the seed area for region growing: select the area in S01 where the mean pixel value Mp(i) is greater than the threshold P t2 At the same time, it is less than P t1 The connected area Sed(i) is sorted in descending order, and the area with the largest pixel mean is selected as the seed area of ​​region growing (RG), that is, Mp(i)<P t1 &&Mp(i)>P t2 is the seed area, where growth begins; t2 It represents the minimum value of the pixel range of the manganese block area, Sed(i) represents the i-th seed area, and Mp(i) represents the pixel mean of the i-th block area.

[0072] In this step, no single pixel is used as a seed point, which avoids scattered points that are not affected by lighting.

[0073] The S2 specifically comprises the following steps:

[0074] S201: In the result Mp(i) calculated in S101, the mean value of the pixels in the region (P t2 ,P t1 ) is used as the seed region of the region growing algorithm;

[0075] S202: In the process of region growing, the pixel is (P t2 ,P t1 ) range as the growth condition. At the same time, if the pixels of two regions intersect, the two regions are merged; that is,

[0076]

[0077] Where x, y represent the coordinates of the pixel, Mk(i) represents the known i-th manganese block area (initially the seed area), It means merging the i-th manganese block area into the j-th manganese block area, and SMk(i) represents the area of ​​the i-th known manganese block area.

[0078] S3, perform morphological opening operation on the grown area to remove thin and long branches and merge cracks, etc. Finally, based on the distance D ij and area S t1 Conduct regional mergers;

[0079] The calculation formula for step S3 is as follows:

[0080]

[0081]

[0082] The opening operation is to first perform the corrosion operation and then the expansion operation. S represents the image area to be processed; A represents the structural element; D ij represents the distance between the centroid of the i-th area and the j-th area, D t1 represents the centroid distance threshold; S t1 It indicates the effective area struck by the mechanical hammer, and Merge means the merge operation.

[0083] S4, locating the manganese block area;

[0084] Mark the merged area according to the extreme value of the pixel point in the area during growth, i.e., min(x i ),max(x i ),min(y i ),max(y i ) to position the rectangular frame. Figure 4 As shown in ; and determine whether the area of ​​the rectangular frame is greater than S t1 If it is greater than, the rectangular frame is divided and the centroids of each rectangular frame are found respectively; the centroids of each rectangular frame are the centroids of the target manganese block area (xi, yi); among them, min(x i ),max(x i ),min(y i ),max(y i ) represents the coordinate extreme values ​​of the pixels identified as manganese blocks in a given area in a two-dimensional plane, i.e., the leftmost, rightmost, bottommost, and topmost four values ​​in the plane.

[0085] The position of the rectangle is to make the mechanical hammer hit the background plate to knock down the manganese block, so the effective range S of the hit is taken into account here. t1 , determine whether the area of ​​the rectangular box is greater than S t1 , if it is greater than, the rectangular frame should be divided and the centroid of each rectangular frame should be found separately.

[0086] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes in form and details may be made without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. A method for identifying manganese block regions using a region growing algorithm based on color features, characterized in that: The steps include: S1. Preprocess the captured images: Divide the image into N blocks and calculate the pixel mean Mp(i) of each block; t1 Determine the reflective area R(i), by the pixel threshold (P t3 ,P t4 ) Determine the black-yellow area B(i); adjust the pixels of the reflective area and the black-yellow area by a color constancy algorithm, and then convert the image into a grayscale image by using a classified grayscale transformation; Among them, P t1 represents the maximum value of the pixel range of the manganese block area; R(i) represents the i-th reflective area; P t3 Indicates the minimum value of pixels in the black and yellow area; P t4 represents the maximum value of the pixels in the black-yellow area; B(i) represents the i-th black-yellow area; S2. Select seed regions for region growing: Select the area pixel mean Mp(i) in S1 that is greater than the threshold P t2 At the same time, it is less than P t1 The connected area Sed(i) is sorted in descending order, and the area with the largest pixel mean is selected as the seed area for region growth, that is, Mp(i)<P t1 &&Mp(i)>P t2 is the seed area, where growth begins; t2 represents the minimum pixel range of the manganese block area, Sed(i) represents the i-th seed area, and Mp(i) represents the pixel mean of the i-th block area; S3, perform morphological opening operation on the grown area to remove the thin and long branches and merge the cracks, and finally calculate the distance D ij and area S t1 Perform region merging, where D t1 represents the centroid distance threshold; S t1 Indicates the effective area struck by the mechanical hammer; S4. Positioning the manganese block area: Mark the merged area according to the extreme value of the pixel point in the area during growth, i.e., min(x i ),max(x i ),min(y i ),max(y i ) to locate the rectangular frame and determine whether the area of ​​the rectangular frame is greater than S t1 If, is greater than, the rectangular frame is divided and the centroids of each rectangular frame are found respectively; the centroids of each rectangular frame are the centroids of the target manganese block area (xi, yi); among them, min(x i ),max(x i ),min(y i ),max(y i ) represents the coordinate extreme value of the pixels identified as manganese blocks in a given area in the two-dimensional plane.

2. The method for identifying manganese block regions using a region growing algorithm based on color features according to claim 1, characterized in that: In S1, the color constancy algorithm is the classified weighted gray world algorithm CWGW.

3. The method for identifying manganese block regions using a region growing algorithm based on color features according to claim 1, characterized in that: The specific steps of S1 are as follows: S101, calculating the average pixel value of the region: Divide the image into N rectangular regions, and calculate the pixel mean Mp(i) of the rectangular region N(i); where N(i) represents the i-th region, and Mp(i) represents the pixel mean of the i-th region; S102, determining whether it is a reflective area or a dark-yellow area; When the pixel mean Mp(i) is greater than the pixel range (Pt2, Pt1) of the manganese block area, that is, greater than the threshold P t1 , then N(i) is considered to be the reflective area R(i); when the pixel mean Mp(i) belongs to the black-yellow area pixel range (P t3 ,P t4 ), then N(i) is considered to be the black-yellow area B(i); if Mp(i) of N(i) is less than the pixel threshold P t1 , then the area is considered to be a manganese block area or other area where the reflection is not serious; where R(i) represents the i-th reflective area, and the manganese block pixel range (P t2 ,P t1 ) is obtained from sample statistics, P t1 Indicates the maximum pixel range of the manganese block, P t2 Indicates the minimum pixel range of the manganese block area; the pixel range of the black and yellow area (P t3 ,P t4 ) is obtained from sample statistics, P t3 Represents the minimum value of pixels in the black and yellow area, P t4 represents the maximum value of the pixels in the black-yellow area; B(i) represents the i-th black-yellow area; S103, using the classified weighted gray world algorithm CWGW to adjust the pixels in the reflective area and the black-yellow area; S104, grayscale conversion of the image: grayscale the reflective area R(i) after the pixels are adjusted in S103 using the average value method, and grayscale the other areas using the weighted average method.

4. The method for identifying manganese block regions using a region growing algorithm based on color features according to claim 3, characterized in that: In S103, the method of using the classified weighted gray world algorithm CWGW to adjust the pixels in the reflective area and the dark yellow area is: 1) Calculate R according to the following formula new , G new , B new : Among them, R new , G new , B new Respectively represent the adjusted pixel R value, G value, and B value; θ is a smaller number used to appropriately increase the values ​​of the R channel and the G channel; R r , G r , B r Respectively represent the R value, G value, and B value of the original image color in the area with severe reflection; 2) R new , G new , B new The pixels are merged to get the pixels after removing the reflection and black-yellow.

5. The method for identifying manganese block regions using a region growing algorithm based on color features according to claim 3, characterized in that: In S104, the formula used to perform grayscale transformation on the image is as follows: Among them, x, y represent pixel coordinates; Gray(x,y) represents the grayscale value at the coordinates (x,y); R(x,y) represents the value of the R channel at the coordinates (x,y); G(x,y) represents the value of the R channel at the coordinates (x,y); B(x,y) represents the value of the R channel at the coordinates (x,y).

6. The method for identifying manganese block regions using a region growing algorithm based on color features according to claim 3, characterized in that: The S2 specifically comprises the following steps: S201: In the result Mp(i) calculated in S101, the mean value of the pixels in the region (P t2 ,P t1 ) is used as the seed region of the region growing algorithm; S202: In the process of region growing, the pixel is (P t2 ,P t1 ) range as the growth condition. At the same time, if the pixels of two regions intersect, the two regions are merged; that is, Where x, y represent the coordinates of the pixel, Mk(i) represents the known i-th manganese block area, It means merging the i-th manganese block area into the j-th manganese block area, and SMk(i) represents the area of ​​the i-th known manganese block area.

7. The method for identifying manganese block regions using a region growing algorithm based on color features according to claim 6, characterized in that: The calculation formula in S3 is as follows: The opening operation is to first perform the corrosion operation and then the expansion operation. S represents the image area to be processed; A represents the structural element; D ij represents the distance between the centroid of the i-th area and the j-th area, D t1 represents the centroid distance threshold; S t1 It indicates the effective area struck by the mechanical hammer, and Merge means the merge operation.

Images