Security and Communication Networks / 2017 / Article / Tab 8

Review Article

A Survey on Breaking Technique of Text-Based CAPTCHA

Table 8

Comparisons of recognition methods.

Recognition methodsMain basisTypical methodsAdvantages Disadvantages

Recognition method based on template matchingGlobal propertyTraversal search matching algorithmThe program is simple and suitable for standard character verification code.The required template library is large; it depends on the choice of template matching.
Local featureShape context matching algorithmThe image information is rich, and it is robust to image scaling and affine transformation.Without rotation invariant.

Recognition method based on character featureCharacter structure featureAlgorithm based on character structure featureSensitive to the details of characters; strong in distinguishing features.The distortion is serious when there are noise interferences.
Character statistical featureAlgorithm based on character statistical featureStrong robustness against noise interference.Targeted; application limited.

Recognition method based on machine learningTemplate matchingSVMStrong approximation ability and generalization ability; good adaptability and high accuracy for small sample space; suitable for two kinds of classification.Not applied to infinite sample space.
KNNIt is better to avoid the problem of imbalanced samples, which is suitable for overlapping samples of the same class.Computation is complex; easy to misjudge in the domain with small sample size.
Traditional methodBPNNFlexible structure design, suitable for multiclass classification.Slow convergence rate; depends on parameters.
Deep learningCNNAccepts an input image directly; automatically extracts features; own robustness to displacement, scale, and deformation; high recognition accuracy.Lack of time dimension; it could not identify using context information.
RNNProcesses data in time series.Time gradient may disappear.
LSTM-RNNOwns the time memory function; effective to prevent gradient disappear.Unable to extract feature automatically.