Brief introduction
Iris recognition method was first proposed by US ophthalmologists Leonard Flom and Arin Safir 1987 years. Iris recognition algorithm is research by Dr. John Dargman out of the University of Cambridge. He proposed a mathematical algorithm to encode the iris, comparison.
Cyclic pupil iris color surrounding tissue, it is rich and varied texture patterns, forms the basis for iris recognition. Iris recognition technology is imaged by scanning an almost infrared light iris patterns, and the degree of similarity is determined by a pattern of the pixel bit XOR operation . Iris recognition process begins to separate the iris image from the eye, then the feature analysis. Theoretically find the probability of two identical irises is 120 parts per million. This is also known to all of biometric technology in the most accurate.
Iris of a person's life change after maturity, and is highly unique, therefore, is a kind of security (close amount) high human biological characteristics. The iris is a different concept of the retina, which is present in the surface of the eye (the lower portion of the cornea), a colored ring around the pupil of the film, not affected by a disease of the eye internal color of the human eye is determined by the iris.
principle
Iris recognition process generally divided into: the iris image acquisition, image preprocessing, feature extraction and feature matching four steps
- Iris image acquisition : the use of specific equipment for the entire digital camera shooting human eye, and the captured image transmitted to the store computer by the image acquisition card.
- Iris image preprocessing : image preprocessing is due to captured eye image includes a lot of redundant information and can not meet the requirements in terms of resolution, etc., need to be pretreated comprises image smoothing, edge detection, image separation operating.
- Iris feature extraction : feature extraction means extracts from the iris image is separated in a unique feature points through a certain algorithm, and encoded.
- Iris feature matching : feature matching means the iris image is encoded in the database of the features previously stored in the pixel is encoded by an exclusive-OR operation for comparison, validation, so as to achieve the purpose of identification.
Iris image acquisition
The use of specific imaging apparatus after obtaining human eye image, acquired picture data, and need to be written to the file in accordance with certain image formats, to complete the required eye image stored in the computer. Most procedure is used BMP format image file, the image data as BMP image files stored without compression, the preprocessing of the image to facilitate future.
Pre-eye image
BMP image mainly 1,4,8,16,24 and 32 other image formats. 32 BMP image 232 represents the image color, the image of each pixel is represented by 32, the toner is not the file format version Generally, 32 to retain the upper 8 bits of the remaining eight bits represent red , 8 bits of green, blue 8-bit representation. 8 BMP image files indicates that the image has 256 colors. Each image pixel is represented by 8 bits, and 8 which find use as an index in the color table, the color of the pixel, 8 BMP images generally also called gray image.
The acquired images are 32-bit color BMP images. 32-bit color image the color image data stored in the large size of the image file is large. However, sometimes the image recognition requirements, these are not necessary, it can be converted to 8-bit grayscale images.
After obtaining the eye image is converted to grayscale image, but also gray image denoising process. The weighted mean of the spatial filtering process, which is a sliding window odd dot on the image slide, the image pixel gray value of the center point of the window corresponding to respective points within the window with the average value of the gradation Alternatively, if the rights specified in the sliding window averaging process of each pixel in the window share of the weight, i.e. the coefficient of each pixel.
Extracting iris image
This process needs to read data of the eye image, an iris image detecting inner and outer edges, the inner round extraction short radius and center coordinates, and then determine the major radius of the iris, the establishment of a polar coordinate system, isolated iris image, the final feature extraction.
And compared to other parts of the eye, the pupil is much smaller gray value, that is, color is much darker, and there is a clear mutation in the gray scale, that is to say other than the gray level of the pupil gray level portion "much darker." Therefore, you can take advantage of this feature.
62 assuming that the image gradation value from the start, and the presence of a plurality of peak points in FIG. We know the darkest colors of the pupil, a peak can be determined for the first gradation pupil distribution. DETAILED first peak was observed, which substantially sinusoidal-like distribution functions, assuming 72 is the peak (value: 884), a left side 62 (value: 0) of the trough, 1/4. 10 period. Accordingly, we determine the right side of the trough 82. Histogram gray-scale image is binarized, a threshold value of 82, the major radius of the iris can be obtained.