Image detection method, image detection apparatus, image processing method, image processing apparatus, and medium

An image processing method comprising the steps of:

• calculating a first characteristic value in accordance with a local image in a first local region at a predetermined position in a search region;
• comparing each second characteristic value previously obtained in accordance with each local image in a plurality of second local regions at a predetermined position in the region of a target image to be detected with said calculated first characteristic value;
• checking whether a second characteristic value meeting a predetermined relation with said first characteristic value is present in said second characteristic values on basis of the result of the comparison;
• obtaining a new position in said search region on the basis of positional information for whole or part of said second local regions when a second characteristic value meeting said relation is present and calculating a new first characteristic value in accordance with the local image in a first local region at said new position;
• judging whether said calculated new first characteristic value meets a predetermined relation with a predetermined second characteristic value among said second characteristic values;
• and
• determining whether said target image is present in said search region on basis of the judgment result.

An image processing method comprising the steps of:

• calculating each first characteristic value in accordance with each local image in a plurality of first local regions in a search region;
• comparing each second characteristic value previously obtained from each local image in a plurality of second local regions at a predetermined position in the region of a target image to be detected with each of said calculated first characteristic values;
• searching a first characteristic value meeting a predetermined relation with said second characteristic values out of said calculated first characteristic values on basis of a result of the comparison and storing the searched first characteristic value and positional information for the local region;
• and
• detecting a target image on basis of said previously obtained second characteristic values and relative positional relation between local regions of the second characteristic values by using the stored first characteristic value and the positional information for its local region as processing objects.

A method for processing an input image comprising the steps of:

• (a) dividing the input image into blocks of pixels;
• (b) calculating a characteristic value for each block formed by step (a);
• (c) storing characteristic values of a reference image;
• (d) comparing a characteristic value calculated in step (b) with a characteristic value stored in step (c);
• and
• (e) recognizing the input image as the reference image based on the comparing of step (d);
• whereby calculating the characteristic value in step (b) includes the steps of:
• (i) summing pixels of at least one color in a block to form an intensity value of the color;
• and
• (ii) forming a vector, at least one dimension of the vector corresponding to the intensity value of the color in the block.