Filtering device and environment recognition system

1. A filtering device, comprising:

an evaluation value deriving module that derives, for a pair of comparison targets having mutual relevance, multiple evaluation values indicative of correlations with any one of extracted parts extracted from one of the comparison targets and multiple extracted parts that are extracted from the other comparison target, respectively;

an evaluation range setting module that sets an evaluation range of the evaluation values, the evaluation range having one of boundaries at the evaluation value with the highest correlation among the multiple evaluation values; and

a difference value determining module that determines whether a difference value for the evaluation value with the highest correlation is valid as the difference value based on the multiple evaluation values and the evaluation range.

2. The filtering device of claim 1 , wherein the difference value determining module determines that the evaluation value with the highest correlation is invalid as the difference value if a ratio of the evaluation values contained in the evaluation range to all the multiple evaluation values is greater than a threshold.

3. The filtering device of claim 1 , wherein the difference value determining module determines that the evaluation value with the highest correlation is invalid as the difference value if an area of a polygon formed with the evaluation values contained in the evaluation range is greater than a threshold, the polygon being formed so as to have the other boundary of the evaluation range as one side of the polygon.

4. The filtering device of claim 1 , wherein the evaluation range setting module changes the evaluation range, according to the evaluation value with the highest correlation.

5. The filtering device of claim 2 , wherein the evaluation range setting module changes the evaluation range according to the evaluation value with the highest correlation.

6. The filtering device of claim 3 , wherein the evaluation range setting module changes the evaluation range according to the evaluation value with the highest correlation.

7. The filtering device of claim 1 , wherein the comparison target is an image and the extracted part is a block consisting of one or more pixels in the image.

8. The filtering device of claim 2 , wherein the comparison target is an image and the extracted part is a block consisting of one or more pixels in the image.

9. The filtering device of claim 3 , wherein the comparison target is an image and the extracted part is a block consisting of one or more pixels in the image.

10. The filtering device of claim 7 , wherein the evaluation range setting module changes the evaluation range according to an average value of luminance of the pixels in a predetermined area of the image.

11. The filtering device of claim 8 , wherein the evaluation range setting module changes the evaluation range according to an average value of luminance of the pixels in a predetermined area of the image.

12. The filtering device of claim 9 , wherein the evaluation range setting module changes the evaluation range according to an average value of luminance of the pixels in a predetermined area of the image.

13. An environment recognition system, comprising:

one or more imaging devices configured to generate a pair of images having mutual relevance;

one or more processors, configured to:

derive, for the pair of generated images, multiple evaluation values indicative of correlations between any one of blocks that is extracted from one of the images and multiple blocks extracted from the other image;

set an evaluation range of the evaluation values, the evaluation range having one of boundaries at the evaluation value with the highest correlation among the multiple evaluation values; and

determine whether a parallax for the evaluation value with the highest correlation is valid as the parallax based on the multiple evaluation values and the evaluation range.