Face recognition apparatus and method using PCA learning per subgroup

A method of extracting a feature vector implemented by a feature vector extracting apparatus comprising a processor having computing device-executable instructions, the method comprising:

• classifying a training data set into a plurality of subgroups, wherein the subgroups include subgroups that are at least classified, from a global data set of the training data, to each have similar characteristic changes and the subgroups further include at least one subgroup that is classified from a local data set of the training data set, the global data set being obtained from a different source than the local data set;
• performing Principal Component Analysis (PCA) learning on each of the subgroups to generate a PCA basis vector set of each of the subgroups;
• projecting the PCA basis vector set of each of the subgroups to the training data set and performing Linear Discriminant Analysis (LDA) learning on a result of the projection to generate a PCA-based LDA (PCLDA) basis vector set of each of the subgroups;
• and
• projecting, by way of the processor, the PCLDA basis vector set of each of the subgroups to an input image and extracting a feature vector set of the input image with respect to each of the subgroups.

A computer-readable recording medium including a program to control an implementation of a method of extracting a feature vector, the method comprising:

• classifying a training data set into a plurality of subgroups, wherein the subgroups include subgroups that are at least classified, from a global data set of the training data, to each have similar characteristic changes and the subgroups further include at least one subgroup that is classified from a local data set of the training data set, the global data set being obtained from a different source than the local data set
• performing Principal Component Analysis (PCA) learning on each of the subgroups to generate a PCA basis vector set of each of the subgroups;
• projecting the PCA basis vector set of each of the subgroups to the training data set and performing Linear Discriminant Analysis (LDA) learning on a result of the projection to generate a PCA-based LDA (PCLDA) basis vector set of each of the subgroups;
• and
• projecting the PCLDA basis vector set of each of the subgroups to an input image and extracting a feature vector set of the input image with respect to each of the subgroups.

A face recognition apparatus comprising:

• a learning unit which performs Principal Component Analysis (PCA) learning on each of a plurality of subgroups constituting a training data set, with the subgroups including subgroups that are at least classified, from a global data set of the training data, to each have similar characteristic changes, and the subgroups further including at least one subgroup that is classified from a local data set of the training data set, the global data set being obtained from a different source than the local data set, the learning unit further then performs Linear Discriminant Analysis (LDA) learning on the training data set, thereby generating a PCA-based LDA (PCLDA) basis vector set of each subgroup;
• a feature vector extraction unit which projects the PCLDA basis vector set of each subgroup to an input image and extracts a first feature vector set of the input image with respect to each subgroup;
• a feature vector storing unit which projects the PCLDA basis vector set of each subgroup to each of a plurality of face images to be registered, thereby generating a second feature vector set of each registered image with respect to each subgroup, and storing the second feature vector set in a database;
• and
• a similarity calculation unit which calculates a similarity between the input image and each registered image.

A face recognition method implemented by a face recognition apparatus comprising a processor having computing device-executable instructions, the method comprising:

• performing Principal Component Analysis (PCA) learning on each of a plurality of subgroups constituting a training data set, with the subgroups including subgroups that are at least classified, from a global data set of the training data, to each have similar characteristic changes, and the subgroups further including at least one subgroup that is classified from a local data set of the training data set, the global data set being obtained from a different source than the local data set, and then performing Linear Discriminant Analysis (LDA) learning on the training data set, thereby generating a PCA-based LDA (PCLDA) basis vector set of each of the subgroups;
• projecting the PCLDA basis vector set of each of the subgroups to an input image and extracting a first feature vector set of the input image with respect to each of the subgroups;
• projecting the PCLDA basis vector set of each of the subgroups to each of a plurality of registered images, thereby generating a second feature vector set of each registered image with respect to each of the subgroups, and storing the second feature vector set in a database;
• and
• calculating, by way of the processor, a similarity between the input image and each registered image.

A computer-readable recording medium including a program to control an implementation of a performing of a face recognition method, the method comprising:

• performing Principal Component Analysis (PCA) learning on each of a plurality of subgroups constituting a training data set, with the subgroups including subgroups that are at least classified, from a global data set of the training data, to each have similar characteristic changes, and the subgroups further including at least one subgroup that is classified from a local data set of the training data set, the global data set being obtained from a different source than the local data set, and then performing Linear Discriminant Analysis (LDA) learning on the training data set, thereby generating a PCA-based LDA (PCLDA) basis vector set of each of the subgroups;
• projecting the PCLDA basis vector set of each of the subgroups to an input image and extracting a first feature vector set of the input image with respect to each of the subgroups;
• projecting the PCLDA basis vector set of each of the subgroups to each of a plurality of registered images, thereby generating a second feature vector set of each registered image with respect to each of the subgroups, and storing the second feature vector set in a database;
• and
• calculating a similarity between the input image and each registered image.

A face recognition method implemented by a face recognition apparatus comprising a processor having computing device-executable instructions, the method comprising:

• performing Principal Component Analysis (PCA) learning on each of subgroups constituting a training data set to generate a PCA basis vector set of each of the subgroups, wherein the subgroups include subgroups that are at least classified from a global data set of the training data to each have similar characteristic changes, and the subgroups further include at least one subgroup that is classified from a local data set of the training data set, the local data set comprising face image data obtained at a same location where face images are recognized;
• projecting the PCA basis vector set of each of the subgroups to the training data set;
• performing Linear Discriminant Analysis (LDA) learning on the training data set resulting from the projection to generate a PCA-based LDA (PCLDA) basis vector set of each of the subgroups;
• and
• extracting, by way of the processor, a feature vector set using the PCLDA basis vector set of each of the subgroups, so that eigen characteristics of each of the subgroups is reflected regardless of a data size of each of the subgroups.