Image processing apparatus, image processing method, and program

An image processing method for use with an image processing apparatus for searching and managing moving image content including:

• image acquiring unit acquiring a moving image content from at least one broadcasting station;
• a storage unit storing a database for registering first layer summary data of a first size from images extracted at a first frequency from plural images making up the moving image content, wherein the first frequency being one of intervals of a predetermined number of frames and at a time of a scene change;
• distance calculation means for calculating a distance between said first layer summary data based on a distance between vectors of elements which are formed by said first layer summary data registered in said database;
• and classification means for clustering into a same class the first layer summary data when the distance calculated by said distance calculation means falls within a predetermined distance, said classification means further clustering the moving image contents into a plurality of classes based on the classes into which said first layer summary data have been clustered;
• said image processing method comprising the steps of:
• causing said distance calculation means to calculate a distance between said first layer summary data based on a distance between vectors of elements which are formed by said first layer summary data registered in said database, wherein each vector being a calculated pixel value average regarding each of the RGB pixels in each region of a partitioned frame image;
• and
• causing said classification means to cluster, into a same class, the first layer summary data when the distance calculated in said distance calculation step falls within said predetermined distance, further causing said classification means to cluster the moving image contents into the plurality of classes based on the classes into which said first layer summary data have been clustered, wherein a maximum of one first layer summary data is clustered into each class.

A program embodied on a non-transitory computer readable medium for use with a computer controlling an image processing apparatus for searching and managing moving image content including:

• image acquiring unit acquiring a moving image content from at least one broadcasting station;
• a storage unit storing a database for registering first layer summary data of a first size from images extracted at a first frequency from plural images making up the moving image content, wherein the first frequency being one of intervals of a predetermined number of frames and at a time of a scene change;
• distance calculation means for calculating a distance between said first layer summary data based on a distance between vectors of elements which are formed by said first layer summary data registered in said database;
• and classification means for clustering into a same class the first layer summary data when the distance calculated by said distance calculation means falls within a predetermined distance, said classification means further clustering the moving image contents into a plurality of classes based on the classes into which said first layer summary data have been clustered;
• said program causing said computer to execute a procedure comprising the steps of:
• causing said distance calculation means to calculate the distance between said first layer summary data based on the distance between vectors of the elements which are formed by said first layer summary data registered in said database, wherein each vector being a calculated pixel value average regarding each of the RGB pixels in each region of a partitioned frame image;
• and
• causing said classification means to cluster, into the same class, the first layer summary data when the distance calculated in said distance calculation step falls within said predetermined distance, further causing said classification means to cluster the moving image contents into the plurality of classes based on the classes into which said first layer summary data have been clustered, wherein a maximum of one first layer summary data is clustered into each class.