# US 2017/0286793 A1 · Method and apparatus for performing registration plate detection with aid of edge-based sliding concentric windows > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2017/0286793 A1 - Document ID: US20170286793A1 - Status: Published application - Kind: A1 - Application number: 15/183,817 - Filed: 2016-06-16 - Published: 2017-10-05 - Earliest claimed: 2016-03-30 - Assignees: Faraday Technology Corp. - Applicants: Faraday Technology Corp. - Inventors: Chien-Chung Tseng - Web page: https://patentgrep.com/patent/US20170286793A1 ## CPC - G06V 20/62: Text, e.g. of license plates, overlay texts or captions on TV images - G06V 10/457: {by analysing connectivity, e.g. edge linking, connected component analysis or slices} - G06V 10/44: Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components - G06V 10/473: {using gradient analysis} - G06V 20/625: {License plates} - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract A method for performing registration plate detection includes: performing image processing on image data of a predetermined region in an image to generate an edge image; and performing registration plate detection according to the edge image and the image data of the predetermined region to determine a location of a registration plate image within the predetermined region, for performing a post-processing corresponding to the registration plate image. The step of performing the image processing includes: performing gradient calculations on image data of the predetermined region to generate a gradient image, where the gradient image includes gradient data of the predetermined region; performing edge threshold estimation on the gradient data of the predetermined region, to update an edge threshold through a set of iterative calculations; and generating the edge image according to the edge threshold and according to selective gradient data of the predetermined region. ## Claims ### Claim 1 A method for performing registration plate detection, the method comprising: performing image processing upon image data of a predetermined region in an input image to generate an edge image, wherein the step of performing the image processing comprises: performing a gradient calculation upon the image data of the predetermined region to generate a gradient image, wherein the gradient image comprises gradient data of the predetermined region; performing edge threshold estimation upon the gradient data of the predetermined region to update an edge threshold via a set of iterative calculations; and processing the image data of the predetermined region according to the edge threshold and according to a selective filter, to generate the edge image; and performing registration plate detection according to the image data of the predetermined region and according to the edge image, to determine a location of a registration plate image in the predetermined region, for performing post-processing corresponding to the registration plate image. ### Claim 2 (depends on 1) The method of claim 1, wherein the gradient image comprises a plurality of pixels corresponding to the predetermined region, and the gradient data of the predetermined region are utilized as gray levels of the plurality of pixels of the gradient image. ### Claim 3 (depends on 2) The method of claim 2, wherein a ratio of extreme pixel represents a ratio of a number of a set of extreme pixels having a common extreme gray level in a binarized version of the gradient image to a total number of pixels in the binarized version; and during the set of iterative calculations, the ratio of extreme pixel approaches a predetermined ratio step by step, and the edge threshold varies with respect to the ratio of extreme pixel. ### Claim 4 (depends on 3) The method of claim 3, wherein the ratio of extreme pixel is determined according to the following equation: R=Gnum/Anum; where the symbol “R” represents the ratio of extreme pixel, the symbol “Gnum” represents the number of the set of extreme pixels, and the symbol “Anum” represents the total number of the pixels in the binarized version. ### Claim 5 (depends on 4) The method of claim 4, wherein the set of iterative calculations is performed according to the following equation: Te(k+1)=Te(k)*(R(k)/Rc)*b; where the symbol “Te” represents a temporary version of the edge threshold, the symbol “k” represents an iteration index of the set of iterative calculations, the symbol “R” represents the predetermined ratio, and the symbol “b” represents a predetermined constant. ### Claim 6 (depends on 3) The method of claim 3, wherein the common extreme gray level is a maximum gray level, each pixel of the set of extreme pixels is a white pixel, and the ratio of extreme pixel is a ratio of white pixels. ### Claim 7 (depends on 1) The method of claim 1, wherein the set of iterative calculations comprises a first iterative calculation; and the step of performing the edge threshold estimation upon the gradient data of the predetermined region to update the edge threshold via the set of iterative calculations further comprises: generating a first estimation edge image according to an initial version of the edge threshold and according to the gradient data of the predetermined region; and performing the first iterative calculation according to the first estimation edge image to generate a first temporary version of the edge threshold, so the edge threshold is updated to be the first temporary version; wherein after the first iterative calculation is finished, at least one subsequent iterative calculation of the first iterative calculation proceeds until at least one predetermined condition corresponding to the set of iterative calculations is satisfied. ### Claim 8 (depends on 7) The method of claim 7, wherein the set of iterative calculations comprises a second iterative calculation; and the step of performing the edge threshold estimation upon the gradient data of the predetermined region to update the edge threshold via the set of iterative calculations comprises: generating a second estimation edge image according to the first temporary version of the edge threshold and according to the gradient data of the predetermined region; and performing the second iterative calculation according to the second estimation edge image to generate a second temporary version of the edge threshold, so the edge threshold is updated to be the second temporary version; wherein after the second iterative calculation is finished, at least one following iterative calculation of the second iterative calculation proceeds until the predetermined condition corresponding to the set of iterative calculations is satisfied. ### Claim 9 (depends on 7) The method of claim 7, wherein when the predetermined condition corresponding to the set of iterative calculations is satisfied, the edge threshold is updated to be a final version of the edge threshold; and the step of generating the edge image comprises: generating the edge image according to the final version of the edge threshold and according to the gradient data of the predetermined region. ### Claim 10 (depends on 1) The method of claim 1, wherein the step of generating the edge image further comprises: utilizing the selective filter to perform selective filtering upon the image data of the predetermined region to generate a selective filtering gradient image, wherein the selective filtering gradient image comprises edge components corresponding to a specific direction; and binarizing the selective filtering gradient image according to the edge threshold to generate the edge image. ### Claim 11 (depends on 10) The method of claim 10, wherein the specific direction is a vertical direction of the input image and not a horizontal direction of the input image; and an edge direction of each edge of a plurality of edges in the edge image approaches the vertical direction and not the horizontal direction. ### Claim 12 (depends on 1) The method of claim 1, wherein the edge image comprises a plurality of pixels corresponding to the predetermined region; and the step of performing the registration plate detection further comprises: performing first sliding concentric window (SCW) calculations upon the plurality of pixels of the edge image, and performing second SCW calculations upon a portion of the predetermined region of the input image to generate an SCW image, wherein the portion of the predetermined region of the input image is selected from the predetermined region according to calculation results of the first SCW calculations, and image data of the SCW image correspond to calculation results of the second SCW calculations. ### Claim 13 (depends on 12) The method of claim 12, wherein the predetermined region of the input image represents a resized image, and the image data of the predetermined region are pixel values of the resized image; and the step of performing the first SCW calculations upon the plurality of pixels of the edge image and performing the second SCW calculations upon the portion of the predetermined region of the input image accordingly to generate the SCW image further comprises: utilizing a plurality of sets of Edge-Based Sliding Concentric Windows (EBSCWs) to calculate edge degrees of the plurality of pixels of the edge image, respectively, wherein the edge degrees represent edge magnitudes of corresponding pixels of the resized image; selecting a set of stronger-edge pixels from the corresponding pixels of the resized image according to the edge degrees, wherein edge magnitudes of where the set of stronger-edge pixels locate are greater than edge magnitudes of where other pixels of the corresponding pixels of the resized image locate; utilizing a plurality of sets of Gray-level-Based Sliding Concentric Windows (GBSCWs) to calculate contrast degrees of the resized image on where the set of stronger-edge pixels locate, respectively, wherein the contrast degrees correspond to contrast magnitudes for where the set of stronger-edge pixels locate, respectively; selecting a set of higher-contrast pixels from the set of stronger-edge pixels according to the contrast degrees, wherein contrast magnitudes for where the set of higher-contrast pixels locate are greater than contrast magnitudes for where other pixels of the set of stronger-edge pixels locate; and setting each pixel of the set of higher-contrast pixels as a predetermined pixel value to generate the SCW image, wherein a pixel value of any pixel of the SCW image, except for the set of higher-contrast pixels, is different from the predetermined pixel value. ### Claim 14 (depends on 13) The method of claim 13, wherein each set of EBSCWs of the plurality of sets of EBSCWs comprises a first inner window and a first outer window; and the step of utilizing the plurality of sets of EBSCWs to calculate the edge degrees of the plurality of pixels of the edge image respectively further comprises: calculating a gray level average of the first inner window and a gray level average of the first outer window, respectively, and performing a first predetermined linear combination on the gray level average of the first inner window and the gray level average of the first outer window, to generate an edge degree of a first pixel of the plurality of pixels of the edge image, wherein the first pixel corresponds to the first inner window and the first outer window. ### Claim 15 (depends on 13) The method of claim 13, wherein each set of GBSCWs of the plurality of sets of GBSCWs comprises a second inner window and a second outer window; and the step of using the plurality of sets of GBSCWs to calculate the contrast degrees of the set of stronger-edge pixels respectively further comprises: calculating a gray level average of the second inner window and a gray level average of the second outer window, respectively, and performing a second predetermined linear combination on the gray level average of the second inner window and the gray level average of the second outer window, to generate a contrast degree of a second pixel of the set of stronger-edge pixels, wherein the second pixel corresponds to the second inter window and the second outer window. ### Claim 16 An apparatus for registration plate detection, comprising: a processing circuit, arranged to control operations of the apparatus, comprising: an image processing module, arranged to process image data of a predetermined region of an input image to generate an edge image, comprising: a gradient image creation unit, arranged to perform a gradient calculation upon the image data of the predetermined region to generate a gradient image, wherein the gradient image comprises gradient data of the predetermined region; and an edge threshold estimation unit, arranged to perform edge threshold estimation upon the gradient data of the predetermined region to update an edge threshold via a set of iterative calculations, wherein the image processing module processes the image data of the predetermined region according to the edge threshold and according to a selective filter, to generate the edge image; and a registration plate detecting module, arranged to perform registration plate detection according to the image data of the predetermined region and according to the edge image, to determine a location of a registration plate image in the predetermined region, for performing post-processing corresponding to the registration plate image. ### Claim 17 (depends on 16) The apparatus of claim 16, wherein the image processing module further comprises: a specific direction edge image creation unit, arranged to process the image data of the predetermined region according to the edge threshold and according to the selective filter to generate the edge image, wherein the specific direction edge image creation unit utilizes the selective filter to perform selective filtering upon the image data of the predetermined region, wherein the selective filtering gradient image comprises edge components corresponding to a specific direction, and the specific direction edge image creation unit binarizes the selective filtering gradient image according to the edge threshold to generate the edge image. ### Claim 18 (depends on 17) The apparatus of claim 17, wherein the specific direction is a vertical direction of the input image and not a horizontal direction of the input image; and an edge direction of each edge of a plurality of edges in the edge image approaches the vertical direction and not the horizontal direction. ### Claim 19 (depends on 16) The apparatus of claim 16, wherein the edge image comprises a plurality of pixels corresponding to the predetermined region; and the registration plate detection module comprises: a Sliding Concentric Window (SCW) image creation unit, arranged to perform a SCW calculation, wherein the SCW image creation unit perform first SCW calculations upon the plurality of pixels of the edge image, and performs second SCW calculations upon a portion of the predetermined region of the input image to generate a SCW image, wherein the portion of the predetermined of the input image is selected from the predetermined region according to calculation results of the first SCW calculations, and image data of the SCW image correspond to calculation results of the second SCW calculations. ### Claim 20 (depends on 19) The apparatus of claim 19, wherein the predetermined region of the input image represents a resized image, and the image data of the predetermined region is pixel values of the resized image; the SCW image creation unit utilizes a plurality of sets of Edge-Based Sliding Concentric Windows (EBSCWs) to calculate edge degrees of the plurality of pixels of the edge image, respectively, wherein the edge degrees represent edge magnitudes of corresponding pixels of the resized image; the SCW image creation unit selects a set of stronger-edge pixels from the corresponding pixels of the resized image according to the edge degrees, wherein edge magnitudes of where the set of stronger-edge pixels locate are greater than edge magnitudes of where other pixels of the corresponding pixels of the resized image locate; the SCW image creation unit utilizes a plurality of sets of Gray-level-Based Sliding Concentric Windows (GBSCWs) to calculate contrast degrees of the resized image on where the set of stronger-edge pixels locate, wherein the contrast degrees correspond to contrast magnitudes for where the set of stronger-edge pixels locate; the SCW image creation unit determines a set of higher-contrast pixels from the set of stronger-edge pixels according to the contrast degrees, wherein contrast magnitudes for where the set of higher-contrast pixels locate are greater than contrast magnitudes for where other pixels of the set of stronger-edge pixels locate; and the SCW image creation unit sets each pixel of the set of higher-contrast pixels as a predetermined pixel value to generate the SCW image, wherein a pixel value of any pixel of the SCW image except the set of higher-contrast pixels is different from the predetermined pixel value.