This blog post clearly examines the technical differences between optical and capacitive fingerprint recognition, explaining how each utilizes light and voltage differences to distinguish ridges and valleys.
As information and communication technologies rapidly advance and services utilizing them become deeply embedded in daily life, incidents of personal information leaks and identity theft are surging. Against this backdrop, public interest in online information security is higher than ever, with biometric recognition technology—considered safer than traditional security methods—garnering particular attention. This is because biometric features like irises, fingerprints, and veins are unique to each individual and, unlike keys or passwords, carry little to no risk of being stolen or replicated. Among the various biometric recognition methods, fingerprint recognition has gained prominence as a representative biometric technology. It is widely used in diverse fields such as door locks, smartphone locks, and e-commerce, owing to its advantages: fingerprint patterns rarely change, and the process is simpler compared to other biometric technologies.
Fingerprint recognition technology is a biometric authentication method that identifies individuals by determining whether a newly acquired fingerprint from a fingerprint input device matches a pre-registered fingerprint. Among the various fingerprint recognition algorithms, the most widely used method is the feature-point-based algorithm. This consists of an extraction process, where feature points—locations where changes occur in the flow of ridges—are extracted from the fingerprint image obtained by the fingerprint sensor to define feature vectors, and a matching process, where the fingerprint recognition result is determined based on these defined feature vectors. Here, a feature vector refers to a set containing various information, such as the type and location of the feature point, and the direction of the ridge where the feature point is located. To perform these steps, a fingerprint image must first be captured via a fingerprint sensor. The defining characteristic of a fingerprint is its composition of ridges (raised areas) and valleys (depressed areas). The sensor utilizes the principle that ridges make direct contact with the sensing surface while valleys do not. It acquires the fingerprint image by analyzing differences in physical quantities, such as light intensity or voltage, between these two regions.
Methods for capturing fingerprint images are categorized into optical fingerprint recognition, which utilizes light intensity, and capacitive fingerprint recognition, which utilizes voltage differences. Optical fingerprint recognition systems employ an illumination device, a prism, and an image sensor. When a finger is placed against the reflective surface of the prism, the ridges directly press against and contact the reflective surface. At this point, moisture or oil present on the ridges forms a thin film on the reflective surface. This thin film refracts or scatters light. When light strikes the reflective surface, the weakened light from the ridges reaches the image sensor, while the light from the valleys, which is not refracted or scattered, reaches the sensor relatively stronger. Ultimately, the image sensor distinguishes ridges from valleys based on these differences in light intensity, converts this into a digital signal, and forms the fingerprint image.
Capacitive fingerprint recognition utilizes a plate densely packed with microscopic capacitive sensors through which a constant current flows. A minute current flows through the skin. When a finger touches the sensor, a voltage difference forms between the finger and the sensor; the greater the distance from the sensor, the larger the voltage difference. This is analogous to water flowing from a higher elevation to a lower one; the greater the height difference, the faster the flow. Consequently, ridge areas have a shorter distance to the sensor, resulting in a smaller voltage difference, while valley areas have a greater distance, yielding a larger voltage difference. This difference is used to obtain the fingerprint image.
Once the fingerprint image is formed, the extraction process proceeds. This process is divided into three stages: line thinning, which adjusts the thickness of the lines composing the fingerprint image to a uniform level below a certain threshold; candidate feature point extraction, which identifies potential feature points from the thinned image; and pseudo-feature point removal, which eliminates fingerprint information that could interfere with final judgment or increase error. During the thinning stage, since noise may be present in the fingerprint image, the image is divided into regions of a fixed size. The ridge flow information within each region is then displayed and simplified into black and white through a binarization process. Subsequently, a smoothing process is applied to make the ridges more clearly visible. After these steps, the ridges and valleys in the fingerprint image are sharply distinguished in black and white.
The clear fingerprint image obtained through the extraction process, i.e., the thinned image, then undergoes candidate feature point extraction to derive information for distinguishing it from other fingerprints. At this stage, based on the thinned ridge information, the system stores data on points where the ridge flow changes, including pseudo-feature points that appear similar to actual feature points due to image distortion, as well as branch points. Since this process may include unnecessary pseudo-features, a separate pseudo-feature removal step is essential. This step involves correcting or partially removing ridges in the image, then compensating for any lost features to extract the final feature set.
Once the fingerprint obtained via the input device is analyzed and the image and features are finalized, the final matching process proceeds. The matching process determines the similarity between the fingerprint image obtained during extraction and the previously registered fingerprint image. This process consists of three steps: finding and aligning corresponding feature points in both fingerprint images, statistically analyzing the coordinates, types, and angles of these corresponding feature points to calculate similarity, and quantifying the calculated similarity to determine the degree of match between the two fingerprints. If the calculated match degree exceeds the threshold value, access is granted; if it falls below the threshold, access is denied.
Fingerprint recognition offers relatively high recognition rates and fast verification speeds. It also has significant advantages in requiring very little space for implementation and having simpler procedures compared to other biometric technologies. Furthermore, its ease of use without burdening users has contributed to its widespread popularity. However, limitations exist: the technology cannot be used if the fingerprint is damaged or the pattern is worn away, and the error rate increases significantly when the finger is sweaty or wet, often preventing recognition. To overcome these shortcomings, continuous technological advancement is essential, including not only sophisticated improvements to existing fingerprint recognition processes but also the introduction of new procedures. As this research and development continues, fingerprint recognition technology will advance toward forms with enhanced reliability and stability.
