# US 10,311,322 B2 · Character information recognition method based on image processing > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,311,322 B2 - Document ID: US10311322B2 - Status: Granted - Kind: B2 - Application number: 15/649,654 - Filed: 2017-07-14 - Published: 2017-10-26 - Granted: 2019-06-04 - Earliest claimed: 2015-01-15 - Assignees: Suntront Tech Co., Ltd - Applicants: Suntront Tech Co., Ltd - Inventors: Jiheng Fei - Web page: https://patentgrep.com/patent/US10311322B2 ## CPC - G06V 30/153: {using recognition of characters or words} - G06F 18/22: Matching criteria, e.g. proximity measures - G06T 11/60: Creating or editing images; Combining images with text - G06V 10/761: {Proximity, similarity or dissimilarity measures} - G06V 30/19013: {Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching (specially adapted for image segmentation G06T7/10; specially adapted for the analysis of motion G06T7/20; specially adapted for image alignment G06T7/30; specially adapted for the calculation of depth from stereo images G06T7/50; specially adapted for position determination G06T7/70)} - G06V 30/196: using sequential comparisons of the image signals with a plurality of references - G06V 30/10: Character recognition - Y02D 10/00: Energy efficient computing, e.g. low power processors, power management or thermal management - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - Y02D: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE ## Abstract The present invention relates to a character information recognition method based on image processing. The method comprises: collecting images to obtain a target character image; then sequentially comparing the target character image with character template images in a character template library to find a maximum of a coincidence area of the character in the target character image with the character templates in the character template images; and when the coincidence area meets a preset condition, determining the target character to be recognized as the character in the corresponding character template image. The character templates are designed to include not only a coincidence-permitted region but also a coincidence-restricted region. The coincidence-restricted region is set, so that the direct comparing and matching of the character templates can be more accurately carried out, thereby improving the recognition speed. ## Claims ### Claim 1 A character information recognition method based on image processing, applicable to a device having a camera module and a computer, wherein the camera module is mounted on a water meter, wherein the character information recognition method comprises the steps of: (a) collecting images of a character wheel of the water meter using the camera module, and obtaining target character images by image preprocessing through the computer, wherein each of the target character images is formed by a background and a target character to be recognized on the background; (b) for one target character image, through the computer, sequentially comparing the target character image with character template images in a character template library to find out a maximum of a coincidence area between the target character in the target character image and a character template in one character template image, and determining the target character to be recognized as a character in a corresponding character template image when the coincidence area meets a preset condition, so as to obtain a number recognition result of the character wheel; and (c) uploading the number recognition result, wherein the character template images within a specified range constitute the character template library; each character template image formed by a background and a character template at a set position on the background is a standard binarized character image; the character template of each character template image comprises a coincidence-permitted region, and the character template of at least one character template image comprises a coincidence-restricted region; the coincidence-permitted region is a set standard character; and the coincidence-restricted region is outside of the coincidence-permitted region and is a key region block for differentiating from the other character templates, wherein the step (b) comprising: for one target character image, while the target character image is sequentially compared with the character template images in the character template library, synchronously calculating a coincidence rate between the target character image and each character template image participating in comparison; and determining the magnitude of the coincidence rate, wherein when the magnitude of the coincidence rate reaches a default condition, the target character to be recognized is recognized as the character of the corresponding character template image, wherein the coincidence rate is positively related to a number of pixels of the target character image coinciding with the coincidence-permitted region, and to a number of pixels of the target character image not coinciding with the coincidence-restricted region. ### Claim 2 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the step (b) comprising: respectively comparing the target character image with all of the character template images in the character template library; directly superposing the target character image on one character template image, and synchronously calculating the coincidence rate between the target character image and the character template image, when the target character image is compared with one character template image; and sequentially obtaining corresponding coincidence rates between the target character image and all of the character template images, and determining the target character to be recognized as the character of the corresponding character template image when fining a maximum coincidence rate. ### Claim 3 (depends on 2) The character information recognition method based on image processing according to claim 2, wherein a coincidence rate calculation formula is: k (a′+b′)/(a+b), the number of pixels of the coincidence-permitted region in the character template image is a, and the number of pixels of the coincidence-restricted region is b; in the process of comparison, the number of the pixels of the target character image coinciding with the coincidence-permitted region is a′, the number of the pixels of the target character image not coinciding with the coincidence-restricted region is b′, and k is a coefficient. ### Claim 4 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the step (b) comprising: respectively comparing the target character image with part of the character template images in the character template library; directly superposing the target character image on one character template image, synchronously calculating the coincidence rate between the target character image and the character template image, and comparing the coincidence rate with a set value when the target character image is compared with one character template image; stopping comparison, and determining the target character to be recognized as the character of the corresponding character template image if the coincidence rate between the target character image and one character template image exceeds the set value. ### Claim 5 (depends on 4) The character information recognition method based on image processing according to claim 4, wherein a coincidence rate calculation formula is: k (a′+b′)/(a+b), the number of pixels of the coincidence-permitted region in the character template image is a, and the number of pixels of the coincidence-restricted region is b; in the process of comparison, the number of the pixels of the target character image coinciding with the coincidence-permitted region is a′, the number of the pixels of the target character image not coinciding with the coincidence-restricted region is b′, and k is a coefficient. ### Claim 6 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the step (b) comprising: respectively comparing the target character image with all of the character template images in the character template library; placing the character template image within a region when the target character image is compared with one character template image, wherein the coverage of the region is larger than the size of the character template image; using the target character image to scan in the region, wherein a coverage of scanning ensures that the target character image covers a part region of the region, where the character template image is located, synchronously calculating the coincidence rate of each scan, and then adopting the maximum value among the coincidence rates calculated each time as the coincidence rate between the target character and the character template image; and sequentially obtaining corresponding coincidence rates between the target character image and all of the character template images, and determining the target character to be recognized as the character of the corresponding character template image when finding a maximum coincidence rate among the corresponding coincidence rates. ### Claim 7 (depends on 6) The character information recognition method based on image processing according to claim 6, comprising a method for generating one character template image for one character which comprises: first shooting an image of the character on the character wheel of the water meter through the camera module, and performing a image processing to obtain a binarized image of the character through the computer; then hollowing the binarized image to generate a boundary through the computer, and drawing a line with a set width in the central portion within the boundary to form the coincidence-permitted region through the computer; then adding the coincidence-restricted region outside the boundary through the computer; and removing the boundary to obtain one of the character template images through the computer. ### Claim 8 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the step (b) comprising: respectively comparing the target character image with all of the character template images in the character template library; placing the target character image within a region when the target character image is compared with one character template image, wherein the coverage of the region is larger than the size of target character image; using the character template image to scan in the region, wherein a coverage of scanning ensures that the character template image covers a part region of the region where the target character image is located, synchronously calculating a coincidence rate of each scan, and then adopting the maximum value among the coincidence rate of each scan as the coincidence rate between the target character and the character template image; and sequentially obtaining corresponding coincidence rates between the target character image and all of the character template images, and determining the target character to be recognized as the character of the corresponding character template image when finding a maximum coincidence rate among the corresponding coincidence rates. ### Claim 9 (depends on 8) The character information recognition method based on image processing according to claim 8, comprising a method for generating one character template image for one character which comprises: first shooting an image of the character on the character wheel of the water meter through the camera module, and performing a image processing to obtain a binarized image of the character through the computer; then hollowing the binarized image to generate a boundary through the computer, and drawing a line with a set width in the central portion within the boundary to form the coincidence-permitted region through the computer; then adding the coincidence-restricted region outside the boundary through the computer; and removing the boundary to obtain one of the character template images through the computer. ### Claim 10 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the step (b) comprising: respectively comparing the target character image with part of character template images in the character template library; placing the character template image in a region when the target character image is compared with one character template image, wherein a coverage of the region is larger than the size of the character template image; using the target character image to scan in the region, wherein a coverage of scanning ensures that the target character image covers a part region of the region where the character template image is located, synchronously calculating the coincidence rate between the target character image and the character template image, and comparing the coincidence rate between the target character image and the character template image with a set value; stopping comparison, and determining the target character to be recognized as the character of the corresponding character template image if the coincidence rate between the target character image and one character template image exceeds the set value. ### Claim 11 (depends on 10) The character information recognition method based on image processing according to claim 10, wherein the coincidence rate is equal to sum of weights of all coinciding points divided by sum of weights of all points in character template; all the coinciding points include coinciding points of the coincidence-permitted region and coinciding points of the coincidence-restricted region; with regard to one point in the character template, the weight of the point in the character template is determined according to whether points which are at the same position as the point are located in the same region in other character templates, wherein the more character templates in which the points at the same position are located in the same region there are, the lower the weight of the point is, otherwise the less character templates in which the points at the same position are located in the same region there are, the higher the weight of the point is, and the same region represents that the points in the other character templates and the point in the character template are located in the coincidence-permitted region at the same time or located in the coincidence-restricted region at the same time. ### Claim 12 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the step (b) comprising: respectively comparing the target character image with part of character template images in the character template library; placing the target character image in a region when the target character image is compared with one character template image, wherein a coverage of the region is larger than the size of the character template image; using the character template image to scan in the region, wherein a coverage of scanning ensures that the character template image covers a part region of the region where the target character image is located, synchronously calculating the coincidence rate between the target character image and the character template image, and comparing the coincidence rate between the target character image and the character template image with a set value; if the coincidence rate between the target character image and one character template image exceeds the set value, then stopping comparison, and determining the target character to be recognized as the character of the character template image. ### Claim 13 (depends on 12) The character information recognition method based on image processing according to claim 12, wherein the coincidence rate is equal to sum of weights of all coinciding points divided by sum of weights of all points in character template; all the coinciding points include coinciding points of the coincidence-permitted region and coinciding points of the coincidence-restricted region; with regard to one point in the character template, the weight of the point in the character template is determined according to whether points which are at the same position as the point are located in the same region in other character templates, wherein the more character templates in which the points at the same position are located in the same region there are, the lower the weight of the point is, otherwise the less character templates in which the points at the same position are located in the same region there are, the higher the weight of the point is, and the same region represents that the points in the other character templates and the point in the character template are located in the coincidence-permitted region at the same time or located in the coincidence-restricted region at the same time. ### Claim 14 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein a coincidence rate calculation formula is: k (a′+b′)/(a+b), the number of pixels of the coincidence-permitted region in the character template image is a, and the number of pixels of the coincidence-restricted region is b; in the process of comparison, the number of the pixels of the target character image coinciding with the coincidence-permitted region is a′, the number of the pixels of the target character image not coinciding with the coincidence-restricted region is b′, and k is a coefficient. ### Claim 15 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the coincidence rate is equal to sum of weights of all coinciding points divided by sum of weights of all points in character template; all the coinciding points include coinciding points of the coincidence-permitted region and coinciding points of the coincidence-restricted region; with regard to one point in the character template, the weight of the point in the character template is determined according to whether points which are at the same position as the point are located in a same region in other character templates, wherein the more character templates in which the points at the same position are located in the same region there are, the lower the weight of the point is, otherwise the less character templates in which the points at the same position are located in the same region there are, the higher the weight of the point is, and the same region represents that the points in the other character templates and the point in the character template are located in the coincidence-permitted region at the same time or located in the coincidence-restricted region at the same time. ### Claim 16 (depends on 15) The character information recognition method based on image processing according to claim 15, wherein the weight of each point of each character template is formed by a cardinal number and an increment, the cardinal number is a fixed value, and the increment is a variable value. ### Claim 17 (depends on 1) The character information recognition method based on image processing according to claim 1, comprising a method for generating one character template image for one character which comprises: first shooting an image of the character on the character wheel of the water meter through the camera module, and performing a image processing to obtain a binarized image of the character through the computer; then hollowing the binarized image to generate a boundary through the computer, and drawing a line with a set width in the central portion within the boundary to form the coincidence-permitted region through the computer; then adding the coincidence-restricted region outside the boundary through the computer; and removing the boundary to obtain one of the character template images through the computer. ### Claim 18 (depends on 1) The character information recognition method based on image processing according to claim 1, wherein the image preprocessing comprises image segmentation, binarization processing, mean filtering, morphological filtering and edge cutting. ### Claim 19 (depends on 1) The character information recognition method based on image processing according to claim 1, comprising a method for generating one character template image for one character which comprises: first shooting an image of the character on the character wheel of the water meter through the camera module, and performing a image processing to obtain a binarized image of the character through the computer; then hollowing the binarized image to generate a boundary through the computer, and drawing a line with a set width in the central portion within the boundary to form the coincidence-permitted region through the computer; then adding the coincidence-restricted region outside the boundary through the computer; and removing the boundary to obtain one of the character template images through the computer.