Image processing method, apparatus and storage medium

• US 7,415,137 B2
• Filed: 11/20/2003
• Issued: 08/19/2008
• Est. Priority Date: 12/13/2002
• Status: Expired due to Fees

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Ishida, Yoshihiro, Ji, Xin, Wang, Libing, Chen, Xinwu
• Primary Examiner(s): Mehta; Bhavesh
• Assistant Examiner(s): Akhavannik; Hadi
• Application Number:
  US10/716,671
• Publication Number:
  US 20040131235A1
• Time in Patent Office: 1,734 Days
• Field of Search: 382115-118, 382/229
• US Class Current: 382/115
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

A method of processing an image includes steps of identifying one candidate for a human face region within an image; calculating a probability that the candidate for human face region represents a human face; and saving the probability as attached information to the image. The method of processing an image can also include steps of identifying one candidate for human face region within an image; calculating a probability that the candidate for human face region represents a human face; judging whether or not the candidate for human face region represents a human face by comparing the probability with a threshold; and saving a result of the step of judging as attached information to the image. According to these methods, results of identifying candidates for human face regions will be saved to the image, and further processes to be conducted on the image can be facilitated.

Claims

1. An image processing method, comprising:

a face probability calculating step of identifying candidates for a human face region within an image and calculating a human face probability that each candidate for the human face region represents a human face;

a portrait probability calculating step of multiplying a value corresponding to one human face probability of one candidate by values of the other human face probabilities of the other candidates and outputting a portrait probability that the image is a portrait; and

a determining step of determining whether the image is the portrait by comparing the portrait probability with a threshold value,

wherein said portrait probability calculating step includes a selecting step of selecting higher K human face probabilities from among the human face probabilities for the candidates calculated in said face probability calculating step and a multiplying step of multiplying a value corresponding to one of the selected K human face probabilities by the values corresponding to the other selected K human face probabilities.

2. An image processing method according to claim 1, further comprising an image processing step of processing the image in accordance with the result in said judging step.

3. An image processing method according to claim 1, further comprising a saving step of saving human face probabilities for the candidates regarding each of a plurality of partial spaces in M-dimensional space,

wherein, in said face probability calculating step, M-dimensional vectors are generated by applying a predetermined algorithm to the candidates for the human face region and a probability that the partial spaces corresponding to the generated M-dimensional vectors represent a human face is calculated from among the saved human face probabilities.

4. An image processing apparatus, comprising:

a face probability calculating unit that identifies candidates for a human face region within an image and calculates a human face probability that each candidate for the human face region represents a human face;

a portrait probability calculating unit that multiplies a value corresponding to one human face probability of one candidate by values of the other human face probabilities of the other candidates and outputting a portrait probability that the image is a portrait; and

a determining unit that determining whether the image is the portrait by comparing the portrait probability with a threshold value,

wherein said portrait probability calculating unit includes a selecting unit that selects higher K human face probabilities from among the human face probabilities for the candidates calculated in said face probability calculating unit and a multiplying unit that multiplies a value corresponding to one of the selected K human face probabilities by the values corresponding to the other selected K human face probabilities.

5. A computer-readable medium encoded with a computer program for performing an image processing method, said method comprising:

a face probability calculating step of identifying candidates for a human face region within an image and calculating a human face probability that each candidate for the human face region represents a human face;

a portrait probability calculating step of multiplying a value corresponding to one human face probability of one candidate by values of the other human face probabilities of the other candidates and outputting a portrait probability that the image is a portrait; and

a determining step of determining whether the image is the portrait by comparing the portrait probability with a threshold values,

wherein said portrait probability calculating step includes a selecting step of selecting higher K human face probabilities from among the human face probabilities for the candidates calculated in said face probability calculating step and a multiplying step of multiplying a value corresponding to one of the selected K human face probabilities by the values corresponding to the other selected K human face probabilities.