Person recognition device and method

1. A person recognition device ( 1 ) for person re-identification in a monitoring region ( 5 ), having a camera apparatus and an evaluation module ( 4 ), wherein the camera apparatus comprises a first camera unit ( 2 ) and a second camera unit ( 3 ), wherein the first camera unit ( 2 ) is configured to record a first monitoring image ( 11 ) of a portion ( 8 ) of the monitoring region ( 5 ) in a visible range and an infrared range, wherein the second camera unit ( 3 ) is configured to record a second monitoring image ( 11 ) of the portion ( 8 ) of the monitoring region ( 5 ) in the visible range, the second camera unit ( 3 ) including a cut-off filter ( 13 ) configured to filter out the infrared range in recording the second image,

wherein the camera apparatus is configured to feed the first monitoring image ( 11 ) and the second monitoring image ( 12 ) to the evaluation module ( 4 ),

wherein the evaluation module ( 4 ) is configured to generate a difference image of the first image and the second image by performing a weighted subtraction between the first monitoring image ( 11 ) and the second monitoring image ( 12 ), the weighted subtraction including transforming identical regions from an object space into identical regions in an image space, re-identify a person ( 9 ) in the monitoring region ( 5 ) based on a difference image of the portion ( 8 ) of the monitoring region ( 5 ) in the infrared range, the difference image being produced from the first monitoring image ( 11 ) and the second monitoring image ( 12 ), and to re-identify the person ( 9 ) based on textile features.

2. The person recognition device ( 1 ) according to claim 1 , wherein a channel of the difference image of the portion ( 8 ) is in a stop band of the cut-off filter.

3. The person recognition device ( 1 ) according to claim 2 , wherein the cut-off filter ( 13 ) forms an infrared cut-off filter and the difference image forms an IR channel image.

4. The person recognition device ( 1 ) according to claim 1 , wherein the camera apparatus forms a stereo camera for stereoscopic evaluation and/or the evaluation module ( 4 ) is configured to generate a stereo image based on the first monitoring image ( 11 ) and the second monitoring image ( 12 ).

5. The person recognition device ( 1 ) according to claim 4 , wherein the evaluation module ( 4 ) is configured to re-identify the person ( 9 ) in the monitoring region ( 5 ) based on the stereo image and/or the stereoscopic evaluation.

6. The person recognition device ( 1 ) according to claim 1 , wherein the monitoring region ( 5 ) and/or the portion ( 8 ) of the monitoring region ( 5 ) forms an expanded near range and/or far range.

7. The person recognition device ( 1 ) according to claim 1 , wherein the evaluation module ( 4 ) is configured to re-identify the person ( 9 ) based on color features.

8. The person recognition device ( 1 ) according to claim 1 , wherein the evaluation module ( 4 ) is configured to segment a person ( 9 ) in the first monitoring image ( 11 ), in the second monitoring image ( 12 ) and/or in a channel image ( 14 ) and to determine a clothing region and re-identify the person ( 9 ) based on the clothing region.

9. The person recognition device ( 1 ) according to claim 1 , wherein the evaluation module ( 4 ) has fabric and/or texture data in a reference database.

10. The person recognition device ( 1 ) according to claim 9 , wherein the fabric and/or texture data comprise information relating to a reflection behavior of the fabric of colors into a stop band of the cut-off filter.

11. The person recognition device ( 1 ) according to claim 1 , wherein the evaluation module ( 4 ) is configured to re-identify the person ( 9 ) based on facial features and the difference image.

12. A method for re-identification of a person ( 9 ) in a monitoring region ( 5 ), wherein a first monitoring image ( 11 ) of a portion ( 8 ) of the monitoring region ( 5 ) is recorded in a visible range and an infrared range using a first camera unit ( 2 ) and a second monitoring image ( 12 ) of the portion ( 8 ) of the monitoring region ( 5 ) is recorded in the visible range using a second camera unit ( 3 ), wherein the second monitoring image ( 12 ) is recorded with a cut-off filter ( 13 ) configured to filtering out the infrared range in recording the second image, wherein a person ( 9 ) that is to be re-identified is determined in the portion ( 8 ) and/or a search for a person ( 9 ) that is to be re-identified is performed in the portion ( 8 ) based on a difference image of the portion ( 8 ) of the monitoring region ( 5 ) in the infrared range, the difference image being produced from the first monitoring image ( 11 ) and the second monitoring image ( 12 ), and wherein the difference image of the first image and the second image is generated by performing a weighted subtraction between the first monitoring image ( 11 ) and the second monitoring image ( 12 ), the weighted subtraction including transforming identical regions from an object space into identical regions in an image space.

13. The method of claim 12 , wherein the person ( 9 ) is located more than twenty meters away from the device ( 1 ).

14. The person recognition device of claim 1 , wherein the person ( 9 ) is located more than twenty meters away from the device ( 1 ).