TWI501166B - A segmented image recognition method and a region identification device thereof - Google Patents

Description

Segmented image recognition method and regional identification device thereof

An image recognition method and an electronic device thereof are particularly related to a segmented image recognition method and a region recognition device thereof.

With the increasing sophistication of technology, the importance of protecting personal data has also increased. The most common way to verify the identity of a user is to enter an account and password. When the entered account and password are correct, the electronic device allows the user to access. In order to ensure the complexity of the password, the general electronic device will now have the shortest length and character type of the password. However, since entering a password is easy to be recorded by a person with a heart, there is still a risk in entering the password. In addition, the user usually remembers not only a set of passwords, but also the dilemma of the user forgetting the set password.

Therefore, in order to reduce the problem of the user's memory password, some manufacturers have proposed using biometrics as a means of identification. For example: iris recognition, voiceprint recognition or fingerprint recognition. Since the cost equipment required for fingerprint recognition is cheaper than other biometrics, and each fingerprint is unique, it is not the same as others, so fingerprint recognition becomes another common means of identification.

In the conventional fingerprint identification method, the fingerprint image of the user is obtained through the image capturing device, and the manner in which the different image capturing device acquires the fingerprint image is also different. The reading method of the Linear Sensor module 110 requires the user to follow the finger. The speed and the pressure channel slide over the line sensing module 110, so that the line sensing module 110 can read the fingerprint image of the user in batches. Please refer to "Figure 1A", which is a schematic diagram of the fingerprint image capture of the line sensing module of the prior art.

For the regional fingerprint image capturing device, the user needs to press a finger on the regional fingerprint image capturing device. The regional fingerprint image capture device reads and identifies the complete fingerprint image at a time. Please refer to FIG. 1B, which is a schematic diagram of fingerprint image capture of a regional fingerprint image capturing device of the prior art. A commonly used regional fingerprint image capturing device 120 is formed by a complementary metal-oxide-semiconductor (CMOS) photosensitive element or a charge-coupled device (CCD).

Various image capturing devices of the prior art have different defects. The line sensing module 110 does not require a large buffer space, but the finger displacement or uneven force is easily generated during the user's pressing of the finger, resulting in an incomplete fingerprint image. Although the regional fingerprint image capturing device 120 can quickly acquire fingerprint images, the regional fingerprint image capturing device 120 requires a large buffer space to store fingerprint images.

If the regional fingerprint image capturing device 120 is used, the manufacturer needs to use a large-capacity buffer space electronic device, which inevitably increases the manufacturing cost of the small embedded device.

In view of the above problems, the present invention provides a segmented image recognition method for a regional fingerprint image capture device having a low storage capacity.

To solve the foregoing problems, the segmented image recognition method disclosed in the present invention includes The following steps: setting a fixed size image selection frame; capturing a fingerprint image block of a corresponding size according to the image selection frame; performing a feature conversion process on the fingerprint image block to generate a corresponding feature block image of the fingerprint image block; Repeat the above steps of capturing and converting until all the feature block images corresponding to the fingerprint image block are obtained; according to the acquisition order of the fingerprint image block, the corresponding feature block image is image-joined to generate a fingerprint image correspondingly The feature image; the fingerprint recognition program for the feature image; and the corresponding fingerprint response device sends a corresponding response message according to the identification result obtained by the fingerprint recognition program.

In addition to the above embodiments, the present invention provides another method for implementing segmented image recognition, which includes the following steps: step a. setting an image selection frame; step b. capturing a corresponding size fingerprint image area according to the image selection frame. Block; step c. normalize the fingerprint image block to obtain the contrast value (Mean) and variance value (Variance) of the fingerprint image block; step d. repeat steps b to c until all fingerprints are recorded The contrast value and the variation value of the image block; step e. Obtain the fingerprint image block from the beginning, and calculate the directional parameter and the mask range of the fingerprint image block; step f. repeat step e. until all the fingerprint images are recorded The directional parameter and the mask range of the block; step g. obtaining the fingerprint image block from the beginning, and normalizing the obtained fingerprint image block and the corresponding contrast value and the variability value to generate the first feature block Image; step h. The first feature block image generates a second feature block image according to the directional parameter and the mask range of the corresponding position; step i. a process of generating a third feature block image; step j. repeating step g to step i until all third feature block images are obtained; step k. performing an image bonding process according to the acquisition order of the third feature block image, Generating a corresponding feature image of the third feature block image; Step 1. Performing a fingerprint identification process on the feature image; Step m. According to the identification result obtained by the fingerprint identification program, the regional fingerprint identification device sends a corresponding response message interest.

In order to realize the operation of the present invention, the present invention also provides a segmented area image recognition device, which comprises an image capturing unit, a storage unit and a processing unit. The image capturing unit is configured to capture the fingerprint image block; the storage unit is configured to store the fingerprint image block, the feature conversion program, the image bonding program and the fingerprint identification program; the processing unit is electrically connected to the image capturing unit and the storage unit; The unit sequentially acquires the fingerprint image block according to the image selection frame through the image capturing unit; the processing unit runs a feature conversion program on the fingerprint image block to generate a feature block image of the corresponding fingerprint image block; the processing unit images the feature block The feature image is obtained through the image bonding process; the processing unit performs a fingerprint recognition process according to the feature image, and returns a corresponding response message.

The invention provides a segmented image recognition method and device thereof, which sequentially captures fingerprint image blocks at different positions through a preset image selection frame. The identification device obtains the corresponding feature block image from the fingerprint image block. The identification device finally joins all of the feature block images to produce a complete feature image. In this way, the identification device can achieve the purpose of fingerprint recognition without setting a large-capacity storage unit, and the user does not need to slide the finger, so that the problem of image misalignment when the pressing force is uneven is not generated.

The features and implementations of the present invention are described in detail below with reference to the drawings.

110‧‧‧Line Sensing Module

120‧‧‧Regional fingerprint image capture device

200‧‧‧Fingerprint identification device

210‧‧‧Image capture unit

220‧‧‧ storage unit

221‧‧‧Fingerprint block

222‧‧‧Characteristic conversion procedure

223‧‧·Image bonding procedure

224‧‧‧Finger identification procedure

230‧‧‧Processing unit

300‧‧‧ Fingerprint image

310‧‧‧Image selection box

510‧‧‧Characteristic block image

511‧‧‧Characteristic block image of multiple fingerprint segments

512‧‧‧Characteristic block images of multiple feature points

520‧‧‧Character Images

FIG. 1A is a schematic diagram of fingerprint image capture of a line sensing module of the prior art.

FIG. 1B is a fingerprint image capture of a regional fingerprint image capture device of the prior art. intention.

Figure 2 is a schematic diagram of the architecture of the present invention.

Figure 3 is a schematic diagram of the overall operational flow of the present invention.

Figure 4A is a schematic diagram of horizontal selection of the image selection frame of the present invention.

Figure 4B is a schematic view of the longitudinal selection of the image selection frame of the present invention.

Figure 4C is a schematic diagram of the oblique selection of the image selection frame of the present invention.

Figure 5 is a schematic view of a feature image joint of the present invention.

Figure 6A is a schematic diagram of the operational flow of an embodiment of the present invention.

Figure 6B shows the image selection frame of the present invention being captured in a non-overlapping manner.

Figure 6C is a feature block image of the present invention having a plurality of feature points.

Figure 6D is a schematic diagram of a feature image of the present invention.

Figure 7A is a schematic diagram of the operational flow of an embodiment of the present invention.

Figure 7B is a schematic diagram of the overlapping selected image selection frame of the present invention.

Figure 7C is a schematic diagram of a feature block image of the fingerprint segment of the present invention.

The 7D figure is a schematic diagram of the feature block image of the feature point of the present invention.

Figure 7E is a schematic diagram of the feature block image of the second feature point of the present invention.

Figure 7F is a schematic diagram of a feature image of the present invention.

FIG. 8A is a schematic flow chart of another fingerprint image recognition method according to the present invention.

FIG. 8B is a schematic diagram of another fingerprint image recognition operation architecture of the present invention.

Figure 9 is a schematic diagram of the identification process applied to other images of the present invention.

Please refer to "Figure 2" for a schematic diagram of the architecture of the present invention. This hair The regional fingerprint identification device (hereinafter referred to as the fingerprint identification device 200) includes an image capturing unit 210, a storage unit 220, and a processing unit 230. The image capturing unit 210 is configured to capture the fingerprint image block 221 . The storage unit 220 is configured to store the fingerprint image block 221, the feature conversion program 222, the image bonding program 223, and the fingerprint recognition program 224. The type of the storage unit 220 may be any one of a random access memory (RAM), a flash memory, or a read only memory (ROM) or a combination thereof.

For example, the built-in programs of the feature conversion program 222, the image bonding program 223, and the fingerprint recognition program 224 can be stored in the flash memory or the read-only memory, and the fingerprint image obtained by the fingerprint recognition device 200 can be Temporarily stored in random access memory. Of course, the above images and various programs can also be stored in the flash memory.

The processing unit 230 is electrically connected to the image capturing unit 210 and the storage unit 220. The image capturing unit 210 is configured to capture the fingerprint image block 221 . It should be noted that, in the fingerprint identification device 200 of the present invention, the user only needs to press the finger on a pressing plate to fix it, and does not need to move the finger to slide the finger on the pressing plate, so that the image capturing unit 210 can be pressed and pressed on the pressing plate. Fingerprint image on. The storage unit 220 is configured to store the fingerprint image block 221, the feature conversion program 222, the image bonding program 223, and the fingerprint recognition program 224.

The processing unit 230 is electrically connected to the image capturing unit 210 and the storage unit 220. The processing unit 230 performs the following identification processing on the captured fingerprint image block 221. Please refer to FIG. 3, which is a schematic diagram of an operational flow of an embodiment of the present invention.

Step S310: setting an image selection frame; step S320: capturing a fingerprint image block of a corresponding size according to the image selection frame; and step S330: performing a feature conversion process on the fingerprint image block to generate a fingerprint image area Blocking the corresponding feature block image; Step S340: repeating steps S320 and S330 until all the feature block images corresponding to the fingerprint image block are obtained; Step S350: according to the acquisition order of the fingerprint image block, the corresponding feature The block image performs an image bonding process to generate a feature image corresponding to the fingerprint image; step S360: performing a fingerprint identification process on the feature image; and step S370: sending a corresponding response message by the fingerprint identification device according to the identification result obtained by the fingerprint identification program .

The fingerprint identification device 200 of the present invention is mainly applied to the small-capacity storage unit 220, and the fingerprint identification device 200 sets an image selection frame 310 (corresponding to step S310), and the size range of the image selection frame 310 is less than one complete Fingerprint image 300. For example, the order of the image selection frame 310 may be selected one by one from top to bottom as in the "Ath 4A" or from left to right in the horizontal direction as in "4B".

The selection method of the image selection frame 310 can be achieved by using the oblique direction selection (zigzag) in addition to the selection of the vertical axis direction or the horizontal axis direction. Please refer to the "4C figure", at the upper left of the "4C figure". It is an image selection frame 310, and the black arrow is the movement path of the image selection frame 310.

Then, the processing unit 230 drives the image capturing unit 210 to sequentially capture different portions of the fingerprint image 300 according to the size of the image selection frame 310. The selected partial fingerprint image is defined as the fingerprint image block 221 . The shooting sequence of the fingerprint image block 221 may be, but is not limited to, a vertical axis direction, a horizontal axis direction, or a diagonal selection, and the image capturing unit 210 is configured for the front and rear two fingerprint image blocks 221 No overlap or partial image overlap The way to shoot.

After acquiring each fingerprint image block 221, the image capturing unit 210 performs a feature conversion process 222 on the fingerprint image block 221 and generates a feature block image of the corresponding fingerprint image block 221. The type of the feature conversion program 222 may be, but not limited to, a binarization program or a feature point conversion program. Of course, a combination of the two may be used. The different feature conversion programs 222 will be described later.

The amount of data of the feature block image generated by the fingerprint image block 221 after being processed by the feature conversion program 222 is smaller than the fingerprint image block 221 . And after the processing unit 230 completes the feature block image, the data of the fingerprint image block 221 is cleared. Therefore, the storage unit 220 does not require a large amount of space to store the captured fingerprint image block 221. The processing unit 230 repeats the steps of capturing the fingerprint image block 221 and the feature conversion program 222 until the entire fingerprint image 300 is captured.

The processing unit 230 performs image bonding processing on all the feature block images 510 in the storage unit 220. The processing unit 230 sequentially joins the feature block images 510 into a feature image 520 of the corresponding complete fingerprint image 300. Please refer to " Figure 5 is a schematic view of a feature image joint of the present invention. The processing unit 230 runs the fingerprint recognition program 224 according to the feature image 520. When the feature image 520 to be recognized matches the fingerprint of the user in the storage unit 220, the processing unit 230 will return a confirmed response message; if the feature image 520 to be recognized does not match the user in the storage unit 220 The processing unit 230 will return an erroneous response message. In addition to the fact that the response message can be played on the screen through the display unit, the corresponding warning sound can also be sent through the speaker.

In the foregoing, the present invention can be implemented according to different feature conversion programs 222. The corresponding processing is performed, and the following describes the binarization program or the feature point conversion program. It is to be noted that the range setting of the image selection frame 310, the selection method of the selection frame, and the combination of the various feature conversion programs 222 described below may be combined and changed depending on the identification device.

The following is a description of the feature point conversion program image, and please refer to "Figure 6A".

Step S610: setting an image selection frame; step S620: capturing a fingerprint image block of a corresponding size according to the image selection frame; step S630: performing a feature point conversion process on the fingerprint image block, and converting the feature block into a corresponding feature block of the plurality of feature points Step S640: Steps S620 and S630 are repeated until all the feature block images corresponding to the fingerprint image block are obtained; Step S650: All feature block images are image-joined to generate corresponding feature images of the fingerprint image. Step S660: performing a fingerprint identification process on the feature image; and step S670: the fingerprint recognition device sends a corresponding response message according to the identification result obtained by the fingerprint identification program.

In order to clearly illustrate this mode of operation, in the present description, a fingerprint image 300 of 100 ^* 40 pixels size is set and an image selection frame 310 of 10 ^* 40 pixels size is set as an illustration, and the image selection frame 310 is in a non-overlapping manner. Make a selection. The processing unit 230 captures the fingerprint image block 221 one by one in a top-down manner in the vertical axis direction. Since the image selection frame 310 is selected in a non-overlapping manner, the processing unit 230 divides the fingerprint image block 221 (100/10=10) into 10 times, as shown in FIG. 6B.

Next, the processing unit 230 performs a feature point conversion process on the fingerprint image block 221, and converts it into a feature block image 510 corresponding to a plurality of feature points (Minutiae), as shown in FIG. 6C, which has multiple Schematic diagram of the feature block image of the feature point. Among them, the fingerprint feature point refers to the intersection of the fingerprint and the break point of the fingerprint. When the processing unit 230 finds the fingerprint feature points, it stores the vector relationship between the feature points and the feature points as the basis for the fingerprint comparison, and the remaining fingerprint data will be discarded at this time. The feature points of the fingerprint image can be generally classified into the following categories: 1. Type Lines 2. Cores and Deltas, and are not limited to what type is used in the present invention. .

After the processing unit 230 completes all the feature block images 510, the processing unit 230 joins all the feature block images 510 and outputs the result of the bonding as the feature image 520. Please refer to the "6D" diagram, which is It is a schematic diagram of the feature image. Finally, the processing unit 230 performs a fingerprint recognition process 224 according to the feature image 520. The processing unit 230 sends a corresponding response message by the fingerprint identification device 200 according to the identification result obtained by the fingerprint recognition program 224.

In addition to the feature conversion program described above, the present invention can be combined with the feature point conversion program and the binarization program as a process. Please refer to FIG. 7A, which is a schematic diagram of the operation flow of another feature conversion program of the present invention. .

Step S710: setting an image selection frame; step S720: capturing a fingerprint image block of a corresponding size according to the image selection frame; step S730: performing a binarization process on the fingerprint image block to generate a feature block image having multiple fingerprint segments Step S740: performing a feature point conversion process on the feature image having the fingerprint line segment, and converting the feature block image into a corresponding plurality of feature points; Step S750: Steps S720 and S740 are repeated until all the feature block images corresponding to the fingerprint image block are obtained. Step S760: All the feature block images are image-joined to generate a feature image corresponding to the fingerprint image. S770: performing a fingerprint identification process on the feature image; and step S780: sending a corresponding response message by the fingerprint identification device according to the identification result obtained by the fingerprint identification program.

In the embodiment, the image selection frame 310 of 10 ^* 40 pixels is set with the fingerprint image 300 of 100 ^* 40 pixels in size, and the image selection frame 310 is captured by using 5 pixels as an overlapping manner. Since the image selection frame 310 is selected in an overlapping manner, the processing unit 230 divides the fingerprint image block 221 (100/10 ^* 2=20) into 20 times, which is shown in Fig. 7B. In the "Fig. 7B", the overlapping selection of the image selection frame 310 can be highlighted. Therefore, in the "Fig. 7B", the continuous image selection frame 310 is represented by a different length, but actually the image selection frame 310 is The size is still the same. Further, in the "Picture 7B", the first image selection frame 310 is displayed with a shorter selection frame, and the second image selection frame 310 is displayed with a longer selection frame, and the third image selection is performed. Block 310 is again displayed in a shorter marquee, analogously.

Since the pitch of the fingerprint line segments in the fingerprint image 300 is too close, it is easy to cause difficulty in feature point identification. Therefore, the fingerprint image 300 can be thinned by the binarization process to generate a clearer fingerprint image 300. After the processing of steps S720 and S730, the feature block image 511 having the multi-fingerprint line segment as in "No. 7C" is obtained. The joint area of the two previous feature block images 511 is determined according to the directivity of the fingerprint line segment of the feature block image 511.

Then, the feature block image 510 after the binarization processing by the processing unit 230 performs a feature point conversion process to generate a feature block image 512 corresponding to the plurality of feature points. Please refer to FIG. 7D. Since the feature block image 510 is obtained by overlapping in this embodiment, the second feature block image 510 (such as "FIG. 7E") will have a partial image overlapping with the previous feature block image 510.

After the processing unit 230 repeatedly performs the above steps, the processing unit 230 will acquire 20 feature block images 510. Since some of the 20 feature block images 510 overlap with the previous/after feature block image 512, the processing unit 230 can also perform image bonding processing according to the alignment of the feature points. Subsequently, the processing unit 230 joins all the feature block images 512, and passes the merged image through the feature point conversion program, and then outputs the image as a feature image 520. Please refer to "FIG. 7F".

Finally, the processing unit 230 performs a fingerprint recognition process 224 according to the feature image 520. The processing unit 230 sends a corresponding response message by the fingerprint identification device 200 according to the identification result obtained by the fingerprint recognition program 224.

In addition to the above-mentioned manner of processing the fingerprint image block 221, the present invention can also capture the same fingerprint image block 221 multiple times, and calculate the corresponding feature parameters each time the fingerprint image block 221 is captured. . Please also refer to "8A" and "8B", which are respectively schematic diagrams of the flow and operation architecture of another fingerprint image recognition method of the present invention.

Step a: setting an image selection frame; step b: capturing a fingerprint image block of a corresponding size according to the image selection frame; step c: normalizing the fingerprint image block to obtain a fingerprint image block Comparison value and variation value; Step d: Repeat steps b to c until all the contrast and variation values of the fingerprint image block are recorded; Step e: Obtain the fingerprint image block from the beginning and calculate the direction of the fingerprint image block Sex parameter and mask range; Step f: Repeat step e. until all the fingerprint image block directional parameters and mask range are recorded; Step g: Obtain the fingerprint image block from the beginning, and the obtained fingerprint image block And normalizing the corresponding comparison value and the variation value to generate the first feature block image; Step h: the first feature block image generates the second feature block according to the directional parameter and the mask range of the corresponding position Image i: performing a binarization process on the second feature block image to generate a third feature block image; step j: repeating step g to step i until all third feature block images are obtained; step k: Performing an image bonding process according to the acquisition order of the third feature block image to generate a feature image corresponding to the third feature block image; Step 1: performing a fingerprint recognition process on the feature image ; And Step m: according to a recognition result obtained by the fingerprint recognition program, a corresponding response message sent by the fingerprint identification device.

Since the fingerprint image selection block 310 of the corresponding size is set to operate in the same manner as described above, the description will not be repeated. In this mode of operation, the same fingerprint image block 221 is captured in different periods, so that the fingers can be captured at different times. The image processing block 221 defines a processing time when different fingerprint images 300 are captured in the same fingerprint image, and defines it as a working round. In other words, different fingerprint images 300 are captured at each turn.

First, during the first operational round, the captured fingerprint image block 221 is normalized to calculate the contrast value (Mean) and the variation value (Variance) of the fingerprint image block 221 . Then, the capture of the different fingerprint image blocks 221 is repeated, and the corresponding contrast value and variation value are calculated. Then, the comparison value and the variation value of all the fingerprint image blocks 221 are recorded into the storage unit.

After the first operational round is completed, the fingerprint image block 221 is captured from the beginning. However, in the first operation round, the directional parameter and the mask range of the fingerprint image block 221 are calculated after the fingerprint image block 221 is acquired in the second operation round. The directional parameter is used to record the directivity of the fingerprint line segment in the fingerprint image block 221, thereby determining the joint area of the two consecutive fingerprint image blocks 221 . The mask range is used to determine the recognition range of the fingerprint image block 221 obtained this time. In other words, in addition to processing the acquired entire fingerprint image block 221, it may also be processed for part of the fingerprint image block 221 .

The process of capturing the fingerprint image block 221 is repeatedly performed until the directional parameter and the mask range of all the fingerprint image blocks 221 are obtained. The directional parameters and the mask range of all the fingerprint image blocks 221 are recorded in the storage unit 220. After all fingerprint image blocks 221 are completed, the second operational round is completed.

At the beginning of the third operational round, the fingerprint image block 221 is also captured from the beginning. Different from the foregoing operation round, the fingerprint image block 221 captured in the third operation round processes the comparison value and the variation value with the corresponding position, and generates the first characteristic block of the position fingerprint image block 221 . image. Next, the first feature block image is according to the corresponding position The directional parameter and the mask range produce a second feature block image. Finally, the second feature block image is binarized to generate a third feature block image. The above steps are repeated until all the fingerprint image blocks 221 are recorded, and the third operational round is completed.

And acquiring the third feature block image obtained by the third operation round according to the acquired sequence, and generating a feature image corresponding to the third feature block image. In this operation, the following processing can also be performed on the feature image: the feature image is binarized, and the feature image 520 having multiple fingerprint segments is generated.

Finally, a fingerprint identification process is performed on the feature image. The processing unit 230 sends a corresponding response message by the fingerprint identification device according to the identification result obtained by the fingerprint identification program.

The present invention provides a segmented fingerprint image recognition method and device thereof, which sequentially captures fingerprint image blocks 221 at different positions through a preset image selection frame 310. The identification device retrieves the corresponding feature block image 510 from the fingerprint image block 221. The identification device finally joins all of the feature block images 510 to produce a complete feature image 520. In this way, the fingerprint identification device 200 can achieve the purpose of fingerprint recognition without setting a large-capacity storage unit 220.

In addition to being applicable to the identification of fingerprint images, the present invention can also be applied to feature recognition of various images, such as text images or face images. Please refer to FIG. 9 , which includes the following steps: Step S910: setting an image selection frame; Step S920: capturing an image block of a corresponding size according to the image selection frame; Step S930: Performing a binary value on the image block a program for generating a corresponding feature block image of the image block; Step S940: Steps S910 and S930 are repeated until the image of the feature block corresponding to all the image blocks is obtained. Step S950: Perform image image matching on the corresponding feature block image according to the order of obtaining the image blocks. Generating the corresponding feature image of the fingerprint image; step S960: performing an image recognition process on the feature image; and step S970: sending a corresponding response message by the image capturing device according to the identification result obtained by the image recognition program.

The setting of the image selection frame and the capturing of the image block are the same as described above. For different images (text images or face images), the binarization program can be used to convert the lines (or points, or blocks) in the image block to be further reduced and skeletonized. After the image. Then, the generated feature block images are combined to generate a feature image. The invention can generate skeletonized or binarized images of corresponding images according to different types of images. The skeletonized images (or binarized images) are compared with the text database for other comparisons.

For example, the present invention enables the personal computer to complete the fingerprint recognition without entering the operating system, and combines the Basic Input/Output System (BIOS) when the computer is turned on, so that the test can be successfully used. The action of booting. In this way, the purpose of data confidentiality can be achieved. Moreover, the user does not need to slide the finger, so there is no problem that the image is misaligned when the pressing force is uneven.

While the present invention has been described above in terms of the preferred embodiments thereof, it is not intended to limit the invention, and the invention may be modified and modified without departing from the spirit and scope of the invention. The patent protection scope of the invention is subject to the definition of the scope of the patent application attached to the specification.

Claims (2)

A segmented image recognition method is applied to an area fingerprint identification device with low storage capacity. The identification method includes the following steps: step a. setting an image selection frame; step b. selecting corresponding according to the image selection frame a fingerprint image block of size; step c. performing a normalization procedure on the fingerprint image block to obtain a comparison value (Mean) and a variation value (Variance) of the fingerprint image block; Step b to step c, until the comparison value and the variation value of all the fingerprint image blocks are recorded; step e. obtaining the fingerprint image block from the beginning, and calculating a directional parameter and a mask of the fingerprint image block Covering the range; step f. repeating step e. until the directional parameter of the fingerprint image blocks and the mask range are recorded; step g. obtaining the fingerprint image block from the head, and the obtained fingerprint image area The block and the corresponding comparison value and the variation value are normalized to generate a first feature block image; step h. the first feature block image is according to the directional parameter of the corresponding position The mask range generates a second feature block image; step i. performing a binarization process on the second feature block image to generate a third feature block image; and step j. repeating step g to step i Until all of the third feature block images are obtained; step k. performing an image bonding process according to the acquisition order of the third feature block images And generating a feature image corresponding to the third feature block image; step 1. performing a fingerprint recognition process on the feature image; and step m. obtaining a recognition result according to the fingerprint identification program, by the regional formula The fingerprint identification device sends a corresponding response message. The segmented image recognition method of claim 1, wherein the generating the feature image in step k further comprises: performing a binarization process on the feature image to generate the feature image having a plurality of fingerprint segments .