Label collection and analysis system based on OCR high-speed image recognition

Abstract

This invention discloses a label acquisition and analysis system based on high-speed OCR image recognition. The system segments a label image into white bar images and black bar images, which are then denoised by a noise reduction unit. The noise-removed white bar and black bar images are merged to obtain a high-resolution image. Multiple line segments within the high-resolution image are extracted, and the system checks whether the splicing points within each line segment belong to the same line equation, generating different processing signals. These signals are then transmitted to an external terminal via an output terminal. This method pre-de-noises the acquired label image, segments it into different images, performs tone conversion to reveal noise, and removes the revealed noise, significantly improving the clarity of the label image. Analyzing the clear label image then enhances the overall analysis effect and avoids misjudgments.

Description

Technical Field

[0001] This invention belongs to the field of label analysis technology, specifically a label acquisition and analysis system based on OCR high-speed image recognition. Background Technology

[0002] OCR refers to software that uses a scanner or digital camera to scan text documents into image files, then analyzes and processes the image files to automatically identify and obtain text information and layout information;

[0003] Patent publication number CN102930264B discloses a product display information collection and analysis system and method based on image recognition technology. The system includes an image acquisition terminal and an image processing center connected to the image acquisition terminal via the Internet. The image processing center includes an image storage server, an image analysis server, a database server, a client, and a router / switch. The image storage server, image analysis server, database server, and client are respectively connected to the router / switch. The method includes: 1) establishing a feature database; 2) acquiring product display images; 3) reading the product display images and analyzing and segmenting them; 4) identifying product information and display location information; 5) identifying the location of price tags and obtaining price tag information; and 6) associating the obtained price tag information with the product barcode. This invention can significantly improve the efficiency of product display information entry and reduce information acquisition costs.

[0004] When collecting and analyzing tags, existing analysis and collection systems match the collected tags with preset tag templates and use matching parameters to determine whether the corresponding tags have defects. However, this analysis and processing method does not remove noise inside the tag image. If there is a lot of noise inside the image, it can easily affect the analysis results and lead to misjudgment, resulting in poor overall tag image detection performance. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a label acquisition and analysis system based on high-speed image recognition of OCR, which is used to solve the technical problem that the noise inside the label image is not removed. If there is a lot of noise inside the image, it will easily affect the analysis results and cause misjudgment, resulting in poor overall label image detection effect.

[0006] To achieve the above objectives, an embodiment of the first aspect of the present invention provides a tag acquisition and analysis system based on high-speed OCR image recognition, comprising a tag acquisition terminal, a processing center, and an output terminal;

[0007] The label acquisition terminal uses OCR high-speed image recognition to acquire the entire label and then sends the acquired label to the processing center.

[0008] The processing center includes an image generation unit, an image segmentation unit, an image analysis unit, and a noise reduction unit;

[0009] The image generation unit converts the collected overall labels into label images and sends the converted label images to the image segmentation unit;

[0010] The image segmentation unit segments the converted label image into a white bar image to be processed and a black bar image to be processed. The segmented white bar image and black bar image are then sent to the denoising unit for denoising processing.

[0011] The denoising unit merges the white bar image to be processed with the pure black template and removes the white noise. It also merges the black bar image to be processed with the pure white template and removes the black noise. The denoising-removed white bar image and black bar image are then sent to the image analysis unit.

[0012] The image analysis unit receives the white bar image and the black bar image after noise removal, merges and analyzes the corresponding white bar image and black bar image after noise removal to obtain a high-definition image, extracts several sets of line segments to be processed from the high-definition image, uses linear equation processing on the line segments to obtain the overall detection result, and outputs the overall detection result through the output terminal.

[0013] Preferably, the image segmentation unit performs the following processing steps to segment the label image:

[0014] S1. Extract the white bar image inside the label image, segment the white bar image from the label image, and set the extracted white bar image as the white bar image to be processed.

[0015] S2. Extract the black bar image inside the label image, segment the black bar image from the label image, and set the extracted black bar image as the black bar image to be processed.

[0016] S3. The white bar image to be processed and the black bar image to be processed are transmitted to the denoising unit, and the denoising unit removes the noise inside the white bar image and the black bar image to be processed.

[0017] S4. The white bar image to be processed and the black bar image to be processed are marked with the same recognition mark, and the white bar image to be processed after removing white noise and the black bar image to be processed after removing black noise are sequentially sent to the image analysis unit.

[0018] Preferably, the step S3 in which the denoising unit removes noise from the white bar image and the black bar image to be processed is as follows:

[0019] S31. Set a set of pure black templates, merge the pure black templates with the white bar image to be processed to obtain the first merged image, then mark the multiple white points inside the first merged image, and the noise reduction unit identifies the marked white points and removes the corresponding white noise.

[0020] S32. Set a set of pure white templates, merge the pure white templates with the black bar image to be processed to obtain a second merged image, then mark multiple black dots inside the second merged image, and the denoising unit identifies the marked black dots and removes the corresponding black noise.

[0021] Preferably, the step of the image analysis unit merging and analyzing the noise-removed white bar image and the black bar image is as follows:

[0022] P1. By identifying the markers, the corresponding white bar images and black bar images to be processed are extracted, and the extracted white bar images and black bar images to be processed are merged to obtain a high-definition image;

[0023] P2. Match the high-resolution image with the preset coordinate template. After matching, obtain several sets of line segments to be processed from the high-resolution image, identifying the start and end points within each line segment. Mark the start point as QS. i Mark the endpoint as MD i , where i represents different line segments to be processed (specifically, a high-definition image is composed of multiple sets of spliced ​​line segments);

[0024] P3. Obtain the starting point QS i Given coordinates (X1, Y1), obtain the endpoint MD. i The coordinates are (X2, Y2);

[0025] P4, Adopt Get the starting point QS i With end point MD i The equation of the line connecting the two points is Y = KX + H;

[0026] P5. Substitute multiple sets of points inside the line segment into the connecting equation Y = KX + H, and check whether the coordinate parameters of the point values ​​conform to the connecting equation Y = KX + H. If they do, generate a normal signal; otherwise, generate a defect signal.

[0027] P6. Then process multiple different line segments in the same way as steps P2-P5 to obtain the corresponding normal signals and defect signals. When there are no defect signals inside the high-definition image, the high-definition image is directly output through the output terminal. When the high-definition image stores defect signals, the high-definition image and defect signals are bundled together and output through the output terminal.

[0028] Preferably, the output terminal outputs high-definition images and high-definition images with defective signals to an external display terminal.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows: the label image is segmented into a white bar image to be processed and a black bar image to be processed. The segmented white bar image and black bar image to be processed are sent to a denoising unit for denoising processing. The denoising unit merges the white bar image to be processed with a pure black template to obtain a first merged image. Multiple white noise points inside the first merged image are removed. The black bar image to be processed is merged with a pure white template to obtain a second merged image. Multiple black noise points inside the second merged image are removed. The white bar image and black bar image to be processed after noise removal are merged to obtain a corresponding high-definition image. Multiple sets of line segments inside the high-definition image are extracted. It is checked whether multiple sets of splicing points inside the line segments belong to the same line equation and different processing signals are generated. The different processing signals are transmitted to an external terminal through an output terminal.

[0030] This method involves pre-processing the acquired label images to remove noise, segmenting the images into different parts, performing tone conversion to reveal the noise, and then removing the revealed noise. This significantly improves the clarity of the label images. Analyzing the clear label images further enhances the overall analysis effect and avoids misjudgments. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the principle framework of the present invention. Detailed Implementation

[0032] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] See also Figure 1 This application provides a label acquisition and analysis system based on high-speed OCR image recognition, including a label acquisition terminal, a processing center and an output terminal;

[0034] The tag acquisition terminal output is electrically connected to the processing center input, and the processing center output is electrically connected to the output terminal input.

[0035] The processing center includes an image generation unit, an image segmentation unit, an image analysis unit, and a denoising unit. The output of the image generation unit is electrically connected to the input of the image segmentation unit. The image segmentation unit and the denoising unit are bidirectionally connected. The output of the image segmentation unit is electrically connected to the input of the image analysis unit.

[0036] The label acquisition terminal uses OCR high-speed image recognition to acquire the entire label and then sends the acquired label to the processing center.

[0037] The image generation unit inside the processing center converts the collected overall labels into label images and sends the converted label images to the image segmentation unit.

[0038] The image segmentation unit segments the converted label image into white bar images and black bar images to be processed. The segmented white bar images and black bar images are then sent to the denoising unit for denoising processing. The specific segmentation steps are as follows:

[0039] S1. Extract the white bar image inside the label image, segment the white bar image from the label image, and set the extracted white bar image as the white bar image to be processed.

[0040] S2. Extract the black bar image inside the label image, segment the black bar image from the label image, and set the extracted black bar image as the black bar image to be processed (specifically, the label is generally in the form of a barcode or QR code, and the barcode or QR code only contains corresponding white or black areas. The white area is the blank area, and the black area is the line segment area).

[0041] S3. The white bar image and the black bar image to be processed are transmitted to the denoising unit. The denoising unit removes noise from the white bar image and the black bar image to be processed. The noise removal steps are as follows:

[0042] S31. Set a set of pure black templates, merge the pure black templates with the white bar image to be processed to obtain the first merged image, then mark the multiple white points inside the first merged image, and the noise reduction unit identifies the marked white points and removes the corresponding white noise.

[0043] S32. Set a set of pure white templates, merge the pure white templates with the black bar image to be processed to obtain a second merged image, then mark multiple black dots inside the second merged image, and the noise reduction unit identifies the marked black dots and removes the corresponding black noise.

[0044] S4. The white bar image to be processed and the black bar image to be processed are marked with the same recognition mark, and the white bar image to be processed after removing white noise and the black bar image to be processed after removing black noise are sequentially sent to the image analysis unit.

[0045] The image analysis unit receives the white bar image and the black bar image to be processed, and merges and analyzes the corresponding white bar image and black bar image by recognizing the tags. The overall detection result of the tag is obtained through the analysis and processing, and the overall detection result is output through the output terminal. The specific method of merging and analyzing is as follows:

[0046] P1. By identifying the markers, the corresponding white bar images and black bar images to be processed are extracted, and the extracted white bar images and black bar images to be processed are merged to obtain a high-definition image (specifically, the high-definition image is the label image after noise removal).

[0047] P2. Match the high-resolution image with the preset coordinate template. After matching, obtain several sets of line segments to be processed from the high-resolution image, identifying the start and end points within each line segment. Mark the start point as QS. i Mark the endpoint as MD i , where i represents different line segments to be processed (specifically, a high-definition image is composed of multiple sets of spliced ​​line segments);

[0048] P3. Obtain the starting point QS i Given coordinates (X1, Y1), obtain the endpoint MD. i The coordinates are (X2, Y2);

[0049] P4, Adopt Get the starting point QS i With end point MD i The equation of the line connecting the two points is Y = KX + H;

[0050] P5. Substitute multiple sets of points inside the line segment into the connecting equation Y = KX + H, and check whether the coordinate parameters of the point values ​​conform to the connecting equation Y = KX + H. If they do, generate a normal signal; otherwise, generate a defect signal.

[0051] P6. Then process multiple different line segments in the same way as steps P2-P5 to obtain the corresponding normal signals and defect signals. When there are no defect signals inside the high-definition image, the high-definition image is directly output through the output terminal. When the high-definition image stores defect signals, the high-definition image and defect signals are bundled together and output through the output terminal.

[0052] External personnel receive the high-definition images and defect signals transmitted by the output terminal through an external display screen, and then process the high-definition images with defect signals in a timely manner to improve the processing effect.

[0053] The data in the above formula are all calculated by removing the dimensions and taking the numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0054] The working principle of this invention is as follows: Tags are pre-collected and sent to a processing center. The tags are converted to generate tag images, which are then sent to an image segmentation unit. The converted tag images are segmented into white bar images and black bar images to be processed. These segmented white bar and black bar images are then sent to a denoising unit for denoising. The denoising unit merges the white bar image with a pure black template to obtain a first merged image. Multiple white noise points within the first merged image are removed. The black bar image is then merged with a pure white template to obtain a second merged image. Multiple black noise points within the second merged image are removed. The noise-removed white bar and black bar images are then merged to obtain a corresponding high-definition image. Multiple line segments within the high-definition image are extracted. It is checked whether multiple splicing points within the line segments belong to the same line equation, and different processing signals are generated. These different processing signals are transmitted to an external terminal through an output terminal.

[0055] This method involves pre-processing the acquired label images to remove noise, segmenting the images into different parts, performing tone conversion to reveal the noise, and then removing the revealed noise. This significantly improves the clarity of the label images. Analyzing the clear label images further enhances the overall analysis effect and avoids misjudgments.

[0056] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A label acquisition and analysis system based on high-speed OCR image recognition, characterized in that, This includes the tag acquisition terminal, processing center, and output terminal; The label acquisition terminal uses OCR high-speed image recognition to acquire the entire label and then sends the acquired label to the processing center. The processing center includes an image generation unit, an image segmentation unit, an image analysis unit, and a noise reduction unit; The image generation unit converts the collected overall labels into label images and sends the converted label images to the image segmentation unit; The image segmentation unit segments the converted label image into a white bar image to be processed and a black bar image to be processed. The segmented white bar image and black bar image are then sent to the denoising unit for denoising processing. The denoising unit merges the white bar image to be processed with the pure black template and removes the white noise. It also merges the black bar image to be processed with the pure white template and removes the black noise. The denoising-removed white bar image and black bar image are then sent to the image analysis unit. The image analysis unit receives the noise-removed white bar image and the black bar image to be processed, merges and analyzes them to obtain a high-definition image, extracts several sets of line segments to be processed from the high-definition image, processes these line segments using linear equations to obtain the overall detection result, and outputs the overall detection result through an output terminal. The steps are as follows: P1. By identifying the markers, the corresponding white bar images and black bar images to be processed are extracted, and the extracted white bar images and black bar images to be processed are merged to obtain a high-definition image; P2. Match the high-resolution image with the preset coordinate template. After matching, obtain several sets of line segments to be processed from the high-resolution image, identifying the start and end points within each line segment. Mark the start point as QS. i Mark the endpoint as MD i , where i represents different line segments to be processed; P3. Obtain the starting point QS i Given coordinates (X1, Y1), obtain the endpoint MD. i The coordinates are (X2, Y2); P4, Adopt Get the starting point QS i With end point MD i The equation of the line connecting the two points is Y = KX + H; P5. Substitute multiple sets of points inside the line segment into the connecting equation Y=KX+H, and check whether the coordinate parameters of the point values ​​conform to the connecting equation Y=KX+H. If they do, generate a normal signal; otherwise, generate a defect signal. P6. Then process multiple different line segments in the same way as steps P2-P5 to obtain the corresponding normal signals and defect signals. When there are no defect signals inside the high-definition image, the high-definition image is directly output through the output terminal. When the high-definition image stores defect signals, the high-definition image and defect signals are bundled together and output through the output terminal.

2. The tag acquisition and analysis system based on high-speed OCR image recognition according to claim 1, characterized in that, The image segmentation unit performs the following processing steps to segment the label image: S1. Extract the white bar image inside the label image, segment the white bar image from the label image, and set the extracted white bar image as the white bar image to be processed. S2. Extract the black bar image inside the label image, segment the black bar image from the label image, and set the extracted black bar image as the black bar image to be processed. S3. The white bar image to be processed and the black bar image to be processed are transmitted to the denoising unit, and the denoising unit removes the noise inside the white bar image and the black bar image to be processed. S4. The white bar image to be processed and the black bar image to be processed are marked with the same recognition mark, and the white bar image to be processed after removing white noise and the black bar image to be processed after removing black noise are sequentially sent to the image analysis unit.

3. The tag acquisition and analysis system based on high-speed OCR image recognition according to claim 2, characterized in that, The step S3 in which the denoising unit removes noise from the white bar image and the black bar image to be processed is as follows: S31. Set a set of pure black templates, merge the pure black templates with the white bar image to be processed to obtain the first merged image, then mark the multiple white points inside the first merged image, and the noise reduction unit identifies the marked white points and removes the corresponding white noise. S32. Set a set of pure white templates, merge the pure white templates with the black bar image to be processed to obtain a second merged image, then mark multiple black dots inside the second merged image, and the denoising unit identifies the marked black dots and removes the corresponding black noise.

4. The tag acquisition and analysis system based on high-speed OCR image recognition according to claim 1, characterized in that, The output terminal outputs high-definition images and high-definition images with defective signals to an external display terminal.

Images