The detection of unique patterns and features in human veins (Palm, and Finger) is called hand vein recognition using matlab. In general, each person has different vein patterns like fingerprint. As well as, this process is also called vascular biometrics authentication. Since it uses blood vessels/veins under the human body’s skin to get a person’s unique identification. Consequently, any electronic reader can get the person’s veins patterns by scanning method.

Mainly, hand vein recognition projects are developed and executed in Matlab. Since this software contains necessary image processing libraries and toolboxes. In specific, PLUS OpenVein Toolkit (1.0.2 version) has the intention to support feature extraction, tracking, matching, and evaluation for hand vein recognition in Matlab.

Further, we have also given you development steps to execute these operations. When you follow these steps, you can construct the whole framework/model for hand vein recognition.

Steps in Hand Vein Recognition using Matlab

• Collect the required raw data
• Perform preprocessing over collected data
• Improve image quality and contrast
• Implement partitioning technique over image
• Execute low-pass filtering method
• Set the adaptive threshold value for normalization
• Perform image thinning for smoothing
• Segment the vein patterns and extract the features on it
• Relate features for same pattern identification
• Classify extracted patterns based on pre-trained data

Now, we can see that the way of handling hand vein images in Matlab software. In this, we have given procedure to create read file, read input image and image search by file format and specific location. Similarly, we also apply other feature extraction and matching operations using suitable functions. Our developers are equipped with legendary skills to handle several varieties of operations in hand vein recognition projects. So, we are adept to overcome any level of complexity in hand vein recognition projects through effective approaches.

How do hand vein images handle in Matlab?

• Make read image file as readImages.m
• Parse the file names in the following order (subject, hand or finger, and sample ID)
• Insert new dataset with appropriate regular expression over parsing filename
• Read the image using Imread () function from the graphics file
  • Syntax – imread(filename,fmt)
  • Read color / grayscale image is specified in filename and image format
  • When a filename is not matched for search, it searches for filename with a format like a filename. fmt.
  • When there is no file in the specified path, it displays the given location
  • The return value of imread function, which is an array-based image where class depends on the file format
  • If the file holds the color image, then enable 3D array similarly if the file holds the gray-scale image, then enable a 2D array

As a new research paradigm, Hand Vein Recognition Using Matlab grabs the attention of the current PhD / MS study community. Through this page, you can learn more technical knowledge on developing hand veins recognition projects!!!

Colormaps for Hand Vein Recognition using Matlab

Next, we can see the support of colormaps in Matlab software. Usually, it prefers three-dimensional arrays (m-by-3 arrays). In this, it supports numbers of double-precision floating-point where the values are between 0 to 1. In general, colormaps are maintained as an integer in many graphics file formats. In the case of Matlab, it converts colormap values while reading of image by imread() function and writing of image by imwrite() function.

To the continuation of colormaps, here we have also mentioned you details of true-color images. When the class of true-color image is double, the range of floating-point is between 0 to 1. When the class of true-color image is uint8, the range of integer is between 0 to 255. When the class of true-color image is uint16, the range of integer is between 0 to 65535. To change the data type of true-color image, you need to rescale data values.

• To transform true color image from uint16 to double
  • RGB64 = double(RGB16)/65535;
• To transform true color image double from to uint16
  • RGB16 = uint16(round(RGB64*65535));
• To transform true color image from uint8 to double
  • RGB64 = double(RGB8)/255;
• To transform true color image double from to uint8
  • RGB8 = uint8(round(RGB64*255));

Now, we can see the important toolbox of Matlab for image processing. Image tool toolbox gives an inter-responsive environment for manipulating, displaying, and navigating within images. By the by, it enables you to perform more image processing operations like acquiring measured distance, pixel value, etc. To start the image tool, imtool () function is used. Further, there are numerous toolboxes and functions to analyze every aspect of the hand vein in the medical image. For your reference, here we have listed the main 3D image analysis function that is commonly used in hand vein recognition using Matlab.

Matlab Functions for Hand Vein Recognition

• dualtree3
  • Used to implement wavelet transform of 3-D dual-tree complex
• wavedec3
  • Used to decompose 3-D wavelet transform
• idualtree3
  • Used to reconstruct 3-D dual-tree complex wavelet transform
• waverec3
  • Used to reconstruct 3-D wavelet transform

All these toolboxes and functions in Matlab are motivated to achieve the best result in pre-processing, feature extraction, and evaluation methods. Since these three main processes are important to accomplish your research goal in hand vein recognition. These processes are collectively executed to extract vein patterns for person identity recognition.

Firstly, pre-processing is used to remove unwanted background and noise information in a scanned input image. And also, it is the first and foremost step in every hand vein recognition and analysis project. Further, it also reduces the dimensionality of input data for fast image processing. Here, we have given you some key pre-processing techniques that are operative for hand vein recognition using Matlab.

Pre-Processing Methods for Hand Vein Recognition

• Circular Shape Basis Gabor Filtering
• Fast and Dynamic Contrast Enhancement
• Masking of Hand Veins
• Even Gabor Filtering for Scattered Elimination
• Emphasis Filtering over Higher Frequency
• Add-on Filtering Methods
  • Unsharp Filter
  • Resizing
  • Wiener Filter
  • Gaussian High Pass Filter
  • Median Filter
• Standardized Compensation of Finger Rotation
• Multiple Channels based Gabor Filter
• Contrast Limited Adaptive Histogram Equalization (CLAHE)
• Gabor Filter / Biological Optical Model for Scattered Light Removal

Secondly, feature extraction is used to extract essential features or patterns in the preprocessed image. Since it mainly focuses on the required processing area by excluding other irrelevant areas in a hand vein image and eye pattern analysis. Below, we have itemized the list of important feature extraction methodologies for hand vein recognition using Matlab.

Feature Extraction Schemes for Hand Vein Recognition

• Principal Curvature
• Gabor Filter
• Phase Correlation
• Wide Line Detector
• Maximum Curvature
• Improved Maximum Curvature
• Local Binary Patterns
• Repeated Line Tracking
• Deformation Data for Finger Vein Recognition
• Speed Up Robust Feature (SURF)
• Scale Invariant Feature Transform (SIFT)
• Isotropic-based Undecimated Wavelet Decomposition
• Post-Processing using Morphological Operations
• Feature Point Matching based on Deformation Tolerant

Thirdly, evaluation metrics are used to assess the performance of the hand vein recognition model. This also brings out the efficiency of the proposed preprocessing, feature extraction, and pattern recognition matching techniques/algorithms. These metrics are not only used to address system performance but also to do a comparative study on existing systems.

Through this study, one can prove that the proposed research work is more reliable and efficient than previous methods. In the following, we have specified a few vital performance parameters that are largely employed for hand vein recognition using Matlab.

Evaluation Metrics for Hand Vein Recognition

• FNMR versus FMR Plot
• EER (same value for FNMR and FMR)
• FMR100 (1% – least FNMR used for FMR)
• ROC, and AUC Plot
• ZeroFMR (0% – least FNMR used for FMR)
• FMR1000 (0.1% – least FNMR used for FMR)

When you are attempting to compute False Non-Match Rate (FNMR) and False Match Rate (FMR), it is necessary to estimate performance in advance. Similarly, other evaluation metrics vary based on your project requirements.

Next, we can see core dataset information for hand vein recognition projects which particularly for Matlab. For any project, the dataset has become an inseparable part to attain your research ambition.

Since, your entire research operations of hand vein recognition like preprocessing, feature extraction, pattern matching, and evaluation are highly dependent on your handpicked dataset. There exist a massive number of datasets that are introduced from different parts of the world. Here, we have given you some latest and significant datasets largely chosen by global developers. Further, we also suggest other datasets based on your project requirements.

Hand Vein Recognition Datasets

Finger Vein Recognition

• SDUMLA-HMT
  • Dataset– Finger Vein Database
  • Category– 100+ subjects (45+ females and 60+ males in 15-30 age)
  • Specification
    • 1 subject – Six finger veins
    • right index, left index, right middle, left middle, right ring, and left ring fingers
    • 1 finger – 6 Samples
    • Database – 3800+ samples in 8 bit-gray-scale and 320 x 440 resolution
  • Purpose– Comparison of RoI segmentation of images
• USM Database
• Dataset– Finger Vein Database
• Category– 120+ subjects (40+ females and 80+ males in 20-50 age)
• Specification
  • 1 subject – Four finger veins
    • right index, left index, right middle, and left middle fingers
  • 1 finger 6 Samples
  • Time interval – above 2 weeks (between sessions)
  • Database – 5900+ samples in 640 x 480 resolution
  • And the depth of the images is 640 x 480 and 8 bit-gray-scale, respectively.
  • Purpose– Comparison of RoI of images
• MMCBNU_6000
• Dataset– Finger Vein Database
• Category– 100+ subjects (15+ females and 80+ males in 15-70 age)
  • Specification
  • 1 subject – Six finger veins
    • right index, left index, right middle, left middle, right ring, and left ring fingers
  • 1 finger – 10 Samples
  • Database – 6000+ samples in 8 bit-gray-scale and 640 x 480 resolution
• Purpose– comparison of RoI of images