Search

CN-121981995-A - Authenticity judging method based on printing ink accumulation uneven characteristics

CN121981995ACN 121981995 ACN121981995 ACN 121981995ACN-121981995-A

Abstract

The invention relates to an authenticity judging method based on printing ink accumulation uneven characteristics, which comprises the steps of collecting forward imaging images and surface fluctuation response enhanced images of a printing ink to be detected, registering the two images to determine a printing effective area, screening an ink accumulation leading area according to gray scale fluctuation or high response in the effective area, positioning micro defect candidate points with partial ink deficiency or gray scale lightening in the leading area, carrying out accumulation backfill consistency test on each candidate point in a neighborhood determined according to resolution or stroke width, wherein the test comprises the steps of judging whether lateral accumulation enhancement formed by extrusion exists on two sides of a defect, judging whether trailing accumulation enhancement consistent with the accumulation direction exists on the downstream of the defect, judging that the accumulation enhancement is related to the space of a defect boundary and is in bypass distribution, and judging as true when the number of the candidate points passing the test reaches a preset threshold, otherwise judging as suspected forging or not judging. The false judgment probability is reduced, and the authenticity confidence with statistical significance can be output.

Inventors

  • CHEN MAN
  • ZHANG BO

Assignees

  • 海南政法职业学院
  • 海南公平司法鉴定中心

Dates

Publication Date
20260505
Application Date
20260126

Claims (10)

  1. 1. The authenticity judging method based on the uneven printing ink accumulation characteristics is characterized by comprising the following steps of: collecting a forward imaging image and a surface relief response enhanced image of a printing text to be detected, and registering the two images to determine an effective area of the printing text; positioning microdefect candidate points with partial ink deficiency or light gray scale in a dominant region, and performing accumulation backfill consistency test on each candidate point in a neighborhood determined according to resolution or stroke width; The inspection comprises the steps of judging whether lateral stacking enhancement formed by extrusion exists on two sides of the defect, judging whether trailing stacking enhancement consistent with the stacking direction exists on the downstream of the defect, judging whether the stacking enhancement on the periphery of the defect is related to the space of the boundary of the defect and is in bypass distribution, and judging that the number of candidate points passing the inspection reaches a preset threshold value, or judging that the candidate points are suspected to be forged or not judging that the candidate points are true.
  2. 2. The method for judging the authenticity based on the uneven printing ink accumulation characteristics according to claim 1, wherein the determination of the effective printing area comprises the steps of binary segmentation of a forward imaging image to obtain a printing foreground, calculation of local relief amplitude in a surface relief response enhanced image, and elimination of pixels with the relief amplitude lower than a preset threshold from the foreground to form the effective printing area.
  3. 3. The method for judging the authenticity based on the printing ink accumulation uneven characteristics according to claim 1, wherein the neighborhood determined according to the resolution or the stroke width is adopted and comprises the steps of carrying out skeleton extraction on strokes in a forward imaging image, estimating the partial stroke width, taking 1-3 times of the partial stroke width as a neighborhood radius of a candidate point, respectively executing accumulation backfill consistency test on at least two different scale neighborhoods, and judging that the candidate point passes the test when the multi-scale test results are consistent.
  4. 4. The method for judging the authenticity based on the non-uniform printing ink accumulation characteristics according to claim 1 is characterized in that the consistency test is carried out by respectively calculating the spatial correlation of lateral accumulation enhancement, trailing accumulation enhancement and bypass distribution as three characteristic scores, carrying out weighted summation according to preset weights to obtain candidate point consistency scores, judging that the candidate points pass the test when the consistency scores are larger than a preset score threshold, and outputting the authenticity confidence based on the number of the candidate points passing the test and the statistics of the consistency scores thereof, wherein the result is used for distinguishing the result as true, suspected counterfeit or non-judgement.
  5. 5. The method for judging the authenticity based on the non-uniform accumulation characteristics of ink in a seal ink according to claim 3, wherein the skeleton extraction is characterized in that a single-pixel skeleton is obtained by refining a stroke prospect, a local stroke width is estimated through distance conversion from a skeleton point to a stroke boundary, the local stroke width is twice as large as the local stroke radius, and when the local stroke width fluctuates in a neighborhood of a candidate point by more than a preset proportion, a median width in the neighborhood is taken as a scale reference of the neighborhood radius of the candidate point.
  6. 6. The method for judging the authenticity based on the uneven printing ink accumulation characteristics according to claim 3, wherein the at least two different scale neighborhoods comprise a first neighborhood and a second neighborhood which are respectively determined by 1 time and 3 times of the width of the partial strokes, and the consistency of the multi-scale test results comprises that candidate points meet the correlation of lateral accumulation enhancement and bypass distribution space in the first neighborhood and the second neighborhood, and the candidate points are judged to pass the test when the trailing accumulation enhancement is met in at least one neighborhood.
  7. 7. The method for judging the authenticity based on the non-uniform stacking characteristics of the ink for the ink jet printer according to claim 3, wherein before the stacking backfill consistency test is executed, a direction self-adaptive neighborhood is constructed according to the tangential direction of a skeleton at a candidate point, the direction self-adaptive neighborhood is an elliptic neighborhood formed by a long axis along the tangential direction of the skeleton and a short axis along the normal direction, the length of the long axis is 2-6 times of the width of a partial stroke, and the length of the short axis is 0.8-2 times of the width of the partial stroke.
  8. 8. The method for judging the authenticity based on the non-uniform stacking features of the ink for the ink marks, according to claim 4, is characterized in that the three feature scores are respectively calculated in at least two scale neighborhoods of candidate points and are normalized and fused to obtain consistency scores, wherein the lateral stacking enhancement scores are obtained by gray gradient or fluctuation response differences at two sides of the defects, the trailing stacking enhancement scores are obtained by unidirectional accumulation enhancement amounts along the trend of strokes, and the bypass distribution space correlation scores are obtained by circumferential continuity of enhancement responses in the neighborhood of the boundary of the defects or along boundary correlation coefficients.
  9. 9. The method for judging the authenticity based on the uneven printing ink accumulation characteristics according to claim 4, wherein the weight and the scoring threshold value are adaptively set according to imaging noise or printing ink contrast, the method comprises estimating background noise variance and stroke contrast in an effective area, and carrying out noise compensation and weight adjustment on the characteristic score according to the background noise variance and the stroke contrast, wherein the authenticity confidence is calculated by the average value and the variance of the number of candidate points passing the inspection and the consistency score.
  10. 10. The method for judging the authenticity based on the non-uniform printing ink accumulation characteristics according to claim 9, wherein the background noise variance is estimated by selecting a flat subarea with the fluctuation response lower than a preset threshold value in an effective area, the stroke contrast is calculated by a stroke internal and external gray level average difference or a local contrast index, and the noise variance and the stroke contrast are mapped together into a noise compensation coefficient for unified normalization of three characteristic scores of lateral accumulation enhancement, trailing accumulation enhancement and bypass distribution space correlation.

Description

Authenticity judging method based on printing ink accumulation uneven characteristics Technical Field The invention relates to the technical field of image processing and computer vision, in particular to an authenticity judging method based on uneven printing ink accumulation characteristics. Background The key problem of the method is that most schemes still stay on the two-dimensional appearance structure or geometric consistency level, and physical mechanism characteristics in the forming process of the seal are difficult to truly describe. Taking CN113705330a as an example, the scheme mainly depends on gray scale length or area comparison in a preset sampling area, and a so-called special point is formed by extracting the boundary point between the seal outline and the background, and then geometric consistency judgment is performed by using the length and the included angle of the connecting line of the special point, and the scheme is still typical image comparison method based on shape, position and geometric relationship. Such methods have a certain distinguishing ability for rough imitation or obviously deformed counterfeit seals under ideal conditions, but are particularly limited in the face of the current high-simulation counterfeit technology. Modern counterfeiting means can enable the overall outline, character morphology, local gray distribution and geometric proportion of a seal to be in high consistency with a true seal through high-precision numerical control engraving, laser engraving or high-resolution printing modes, so that statistical results based on gray length, area or random dot-taking connecting lines of sampling rings are very easy to fall into a preset threshold range, and misjudgment is generated. The method is highly sensitive to seal alignment, size normalization and center positioning, and once slight rotation errors, stretching distortion, paper wrinkles or scanning resolution changes exist in the imaging process, linkage influence can be generated on the position and geometric calculation result of a sampling area, so that the discrimination result is unstable. The proposal completely ignores the phenomenon of uneven microscopic accumulation of the ink in the real imprinting process due to stress, friction and material rheology, only regards the imprinting as a static two-dimensional graphic object, and cannot distinguish the real imprinting which looks the same but has completely different formation mechanisms from mechanical copying or printing of the imitated imprinting, which is also the root cause of insufficient reliability in judicial identification and high-risk application scenes. In contrast, CN110728277a introduces the more advanced computer vision and deep learning ideas of feature map, candidate frame, text region leveling and OCR recognition, but the technical goal of this scheme is mainly focused on stamp detection, positioning and text content recognition, instead of stamp authenticity discrimination, whose core focus is where the stamp is written on, what is written on the stamp, instead of whether the stamp is formed by real stamping or not. The method relies on the feature map and the candidate frame classification result, is greatly influenced by training data distribution, network generalization capability and seal style diversity, and the output result is mainly position and text information, and lacks direct depicting capability for differences of true and false seals on a physical forming mechanism. In the scene of authenticity discrimination, even if character recognition is completely correct and position detection is accurate, the possibility of forging a seal or synthesizing a seal at a later stage with high quality cannot be excluded. In addition, the method is generally sensitive to imaging conditions, illumination changes, inkpad aging and paper background, and once the model is out of the training distribution environment, the false detection and false detection probability can rise significantly. Disclosure of Invention The invention aims to provide an authenticity judging method based on uneven printing ink accumulation characteristics, so as to solve part of defects and shortcomings pointed out in the background art. The method for judging the authenticity based on the uneven printing ink accumulation characteristic comprises the following steps of collecting a forward imaging image and a surface fluctuation response enhanced image of a printing text to be detected, registering the two images to determine a printing text effective area, screening an ink accumulation leading area according to gray scale fluctuation or height response in the effective area, and determining a printing text effective area according to the gray scale fluctuation or height response; positioning microdefect candidate points with partial ink deficiency or light gray scale in a dominant region, and performing accumulation backfill consistency test on each candid