Academia.eduAcademia.edu

Figure 3 – uploaded by MOHD RAZIF SHAMSUDDIN

See full PDFdownloadDownload figure
In figure 4(a), we can distinguish the angle of the stroke of the signature slope is inclining leftward thus making this signature categorized as leftward slant. Fig. 4: Signature left slant (a), vertical slant (b), right slant (c)

Figure 4 In figure 4(a), we can distinguish the angle of the stroke of the signature slope is inclining leftward thus making this signature categorized as leftward slant. Fig. 4: Signature left slant (a), vertical slant (b), right slant (c)

Related Figures (18)

However, this paper will represent on-line signature identification. In on-line signature identification a tablet instrument are available in order to extract the signature information more accurately. Signature data can be captured on-line in x and y values in a form of continuous coordinates. These continuous coordinates will be mapped into a grid the need of the system. Refer to figure 1 for the offline signature capturing flow.
However, this paper will represent on-line signature identification. In on-line signature identification a tablet instrument are available in order to extract the signature information more accurately. Signature data can be captured on-line in x and y values in a form of continuous coordinates. These continuous coordinates will be mapped into a grid the need of the system. Refer to figure 1 for the offline signature capturing flow.
Fig. 6: Raw data captured from tablet Figure 6 shows how the raw data is captured using the tablet. As the pen moves toward the right, the x point value will increase. Meanwhile if the pen moves in a higher position, the y value will decrease. The wide size of the tablet allows freedom of creating the signature by providing enough space for the signer. Data are captured in y and x position and mapped into grid in sequence. The next step is to group all significant points that can form one stroke. Next, all possible strokes will be accumulated and categorized.
Fig. 6: Raw data captured from tablet Figure 6 shows how the raw data is captured using the tablet. As the pen moves toward the right, the x point value will increase. Meanwhile if the pen moves in a higher position, the y value will decrease. The wide size of the tablet allows freedom of creating the signature by providing enough space for the signer. Data are captured in y and x position and mapped into grid in sequence. The next step is to group all significant points that can form one stroke. Next, all possible strokes will be accumulated and categorized.
Fig. 7 Filtering stroke points 4.1.3 Feature Extraction a d e After the filtering steps, slant points are extracted from the accumulated slants list during the filtering stage in a form of x- and y-axis values. Two mathematical function based on the value of first and ast point of extracted slant. The angle wil right slant or upright slant (vertical). From the angle, function of (3). The entire accumulated slant list wil be calculated for its angle and degree. The slant wil lgorithms were proposed in this research to extrac he slant features. The first feature to be extracted is he angle of the slant which is determined by etermine whether the slant attributes is left slant, he degree can be acquired by a mathematica hen be categorized for their attributes as explained arlier. The dominating attributes will be selected as he global slant of the signature that will determine he overall signature slant.
Fig. 7 Filtering stroke points 4.1.3 Feature Extraction a d e After the filtering steps, slant points are extracted from the accumulated slants list during the filtering stage in a form of x- and y-axis values. Two mathematical function based on the value of first and ast point of extracted slant. The angle wil right slant or upright slant (vertical). From the angle, function of (3). The entire accumulated slant list wil be calculated for its angle and degree. The slant wil lgorithms were proposed in this research to extrac he slant features. The first feature to be extracted is he angle of the slant which is determined by etermine whether the slant attributes is left slant, he degree can be acquired by a mathematica hen be categorized for their attributes as explained arlier. The dominating attributes will be selected as he global slant of the signature that will determine he overall signature slant.
Fig. 7 Eight possible angle of slant From the acquired two dimensional series of x and y data, the possible slant is identified starting from the pen touches the tablet to when the pen is lifted up or when the direction of the slant changes between two points. Slant direction has been identified to be 8 possible directions as shown in figure 7 All these directions would be computed for the difference value of current point and the previous point:
Fig. 7 Eight possible angle of slant From the acquired two dimensional series of x and y data, the possible slant is identified starting from the pen touches the tablet to when the pen is lifted up or when the direction of the slant changes between two points. Slant direction has been identified to be 8 possible directions as shown in figure 7 All these directions would be computed for the difference value of current point and the previous point:
A number of direction change can be accepted due to slant of ‘C’, inverted ‘C’ or ‘S’ shaped strokes which is a bit curvy. The direction changes that are accepted as one stroke is as depicted in table 2.
A number of direction change can be accepted due to slant of ‘C’, inverted ‘C’ or ‘S’ shaped strokes which is a bit curvy. The direction changes that are accepted as one stroke is as depicted in table 2.
After the filtering process is finished, the entire slant stroke is calculated for its angle and degree. The calculation of degree is made using the equation (4). Then, the angle and degree of the stroke will be classified to its category. The process will repeat itself until the entire filtered stroke is determined for its degree and categorized to its group. Finally, the algorithm will check the most dominant category of the slant in order to classify the overall signature slant category.
After the filtering process is finished, the entire slant stroke is calculated for its angle and degree. The calculation of degree is made using the equation (4). Then, the angle and degree of the stroke will be classified to its category. The process will repeat itself until the entire filtered stroke is determined for its degree and categorized to its group. Finally, the algorithm will check the most dominant category of the slant in order to classify the overall signature slant category.
Fig. 10: Step 2 Process Flow The acquired raw data of x and y values is used to represent the signature form. As the x values increases, it shows the movement of the signature stroke from left to the right. Meanwhile as the y values increases, it shows the signature stroke is moving downward. This happens as the data representation of the tablet recorded a higher y value as the stroke moves to the bottom part of the tablet.
Fig. 10: Step 2 Process Flow The acquired raw data of x and y values is used to represent the signature form. As the x values increases, it shows the movement of the signature stroke from left to the right. Meanwhile as the y values increases, it shows the signature stroke is moving downward. This happens as the data representation of the tablet recorded a higher y value as the stroke moves to the bottom part of the tablet.
Fig. 11: Step 3 Process Flow In the third step, the entire filtered slant list is calculated for its angle and degree. The process will end when the entire stroke is determined for its angle and degree.
Fig. 11: Step 3 Process Flow In the third step, the entire filtered slant list is calculated for its angle and degree. The process will end when the entire stroke is determined for its angle and degree.
This is followed by the final step. In the final step, the stroke is classified for its slant category and
This is followed by the final step. In the final step, the stroke is classified for its slant category and
Table 3: Result 1
Table 3: Result 1
Table 4: Changes in Result 2 The results have shown an increasing number of strokes detected in every signature. Some strokes that is not identifiable before is attained and classified. As we can see signature number 19 slants is identifiable when the range of rule is changed. However, the new results have affected several slant classification change from the earlier results. We can see the changes in signature 50. This may due to the increasing number of stokes identified. The algorithm is then tested with 10 created proofed
Table 4: Changes in Result 2 The results have shown an increasing number of strokes detected in every signature. Some strokes that is not identifiable before is attained and classified. As we can see signature number 19 slants is identifiable when the range of rule is changed. However, the new results have affected several slant classification change from the earlier results. We can see the changes in signature 50. This may due to the increasing number of stokes identified. The algorithm is then tested with 10 created proofed
Table 5: 10 Predefined proofed samples.
Table 5: 10 Predefined proofed samples.
Table 6: Result of 10 created proofed samples
Table 6: Result of 10 created proofed samples
Related topics:
MathematicsComputer ScienceSignature RecognitionOnline Signature VerificationDegree in music
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved