Method and apparatus for detecting structural elements of subjects

A detecting apparatus, comprising:

• target region setting means, for setting judgment target regions within an image;
• and
• judging means, for judging whether images within the judgment target regions set by the target region setting means represent detection targets, based on a plurality of characteristic amounts, which are calculated from pixel values of pixels that constitute the images within the judgment target regions;
• the judging means executing the processes of:
• classifying and recording formulas for calculating the plurality of characteristic amounts into N groups (N>1), such that the amount of calculations required to obtain characteristic amounts by formulas in a Kth group (1≦K<N) is less than the amount of calculations required to obtain characteristic amounts by formulas in a (K+1)th group;
• outputting judgment results indicating that an image within a judgment target region does not represent the detection targets, in the case that it is judged that the image within the judgment target region does not represent the detection targets, based on characteristic amounts obtained by formulas in a Kth group;
• performing judgment, based on characteristic amounts obtained by formulas in a (K+1)th group, in the case that it is judged that an image within a judgment target region represents the detection targets, based on characteristic amounts obtained by formulas in a Kth group;
• and
• outputting judgment results indicating that the image within the judgment target region represents the detection targets, in the case that it is judged that the image within the judgment target region represents the detection targets, based on characteristic amounts obtained by formulas in the Nth group.

A structural element detecting method for detecting predetermined structural elements included in predetermined subjects from within images that include the subjects, comprising:

• a first structural element candidate detecting process, for detecting structural element candidates by discriminating characteristics inherent to the predetermined structural elements, from among characteristics of the images;
• a first subject detecting process, for detecting subjects that include the structural element candidates by discriminating characteristics inherent to the predetermined subjects, from among characteristics of the images in the vicinities of the structural element candidates;
• and
• a structural element confirming process, for confirming that the structural element candidates are the predetermined structural elements, by discriminating characteristics inherent to the predetermined structural elements with a higher degree of accuracy than that of the structural element candidate detecting section, from among the characteristics of the images in the vicinities of the structural element candidates, which are included within the subjects detected by the subject detecting section;
• wherein the structural element confirming process performs discrimination, utilizing data regarding subjects detected by the subject detecting process;
• wherein the number of structural element candidates, detected by a first structural element candidate detecting process and confirmed as being the predetermined structural elements, is compared against the maximum number of structural elements which are possibly included within the subjects detected by the subject detecting process;
• a second structural element candidate detecting process, having a higher detection rate than the first structural element candidate detecting process, is executed in the case that number of structural element candidates, which have been detected by the first structural element candidate detecting process and confirmed to be structural elements by the structural element confirming section, is less than the maximum number;
• and
• the structural element confirming process is executed with respect to the structural element candidates which have been detected by the second structural element candidate detecting process,
• wherein at least one of the first structural element candidate detecting process, the first subject and second detecting processes, and the a structural element confirming process are performed by a processor.

A detecting method, comprising:

• a target region setting step, for setting judgment target regions within an image;
• and a judging step, for judging whether images within the judgment target regions set in the target region setting step represent detection targets, based on a plurality of characteristic amounts, which are calculated from pixel values of pixels that constitute the images within the judgment target regions;
• the judging step executing the processes of:
• classifying and recording formulas for calculating the plurality of characteristic amounts into N groups (N>1), such that the amount of calculations required to obtain characteristic amounts by formulas in a Kth group ((1 ≦K<N) is less than the amount of calculations required to obtain characteristic amounts by formulas in a (K+1)th group;
• outputting judgment results indicating that an image within a judgment target region does not represent the detection targets, in the case that it is judged that the image within the judgment target region does not represent the detection targets, based on characteristic amounts obtained by formulas in a Kth group;
• performing judgment, based on characteristic amounts obtained by formulas in a (K+1)th group, in the case that it is judged that an image within a judgment target region represents the detection targets, based on characteristic amounts obtained by formulas in a Kth group;
• and
• outputting judgment results indicating that the image within the judgment target region represents the detection targets, in the case that it is judged that the image within the judgment target region represents the detection targets, based on characteristic amounts obtained by formulas in the Nth group;
• wherein at least one of the target region setting and judging is performed by a processor.

A computer readable recording medium having a program that provides a function of detecting predetermined structural elements included in predetermined subjects from within images that include the subjects, the program causing a computer to execute:

• first structural element candidate detecting process, for detecting structural element candidates by discriminating characteristics inherent to the predetermined structural elements, from among characteristics of the images;
• a first subject detecting process, for detecting subjects that include the structural element candidates by discriminating characteristics inherent to the predetermined subjects, from among characteristics of the images in the vicinities of the structural element candidates;
• and
• a structural element confirming process, for confirming that the structural element candidates are the predetermined structural elements, by discriminating characteristics inherent to the predetermined structural elements with a higher degree of accuracy than that of the structural element candidate detecting section, from among the characteristics of the images in the vicinities of the structural element candidates, which are included within the subjects detected by the subject detecting section;
• wherein the structural element confirming process performs discrimination, utilizing data regarding subjects detected by the subject detecting process;
• wherein the number of structural element candidates, detected by a first structural element candidate detecting process and confirmed as being the predetermined structural elements, is compared against the maximum number of structural elements which are possibly included within the subjects detected by the subject detecting section;
• a second structural element candidate detecting process, having a higher detection rate than the first structural element candidate detecting process, is executed in the case that number of structural element candidates, which have been detected by the first structural element candidate detecting process and confirmed to be structural elements by the structural element confirming section, is less than the maximum number;
• and
• the structural element confirming process is executed with respect to the structural element candidates which have been detected by the second structural element candidate detecting process.

A computer readable recording medium having a program that provides a function of detecting detection targets included in judgment target regions, the program causing a computer to execute:

• a target region setting step, for setting the judgment target regions within an image;
• and
• a judging step, for judging whether images within the judgment target regions set in the target region setting step represent detection targets, based on a plurality of characteristic amounts, which are calculated from pixel values of pixels that constitute the images within the judgment target regions;
• the judging step executing the processes of:
• classifying and recording formulas for calculating the plurality of characteristic amounts into N groups (N>1), such that the amount of calculations required to obtain characteristic amounts by formulas in a Kth group (1≦K<N) is less than the amount of calculations required to obtain characteristic amounts by formulas in a (K+1)th group;
• outputting judgment results indicating that an image within a judgment target region does not represent the detection targets, in the case that it is judged that the image within the judgment target region does not represent the detection targets, based on characteristic amounts obtained by formulas in a Kth group;
• performing judgment, based on characteristic amounts obtained by formulas in a (K+1)th group, in the case that it is judged that an image within a judgment target region represents the detection targets, based on characteristic amounts obtained by formulas in a Kth group;
• and
• outputting judgment results indicating that the image within the judgment target region represents the detection targets, in the case that it is judged that the image within the judgment target region represents the detection targets, based on characteristic amounts obtained by formulas in the Nth group.