Intelligent zoom method and video system implementing same
A video system is provided. The video system, comprising: a plurality of imaging sensors; a plurality of image signal processors, each associated with an imaging sensor; an image stitching engine to stitch images output by the image signal processors; a zoom engine configured to determine a region of interest for a zoom function wherein said region of interest includes a left edge and a right edge, wherein said left edge is set based on coordinates associated with a left most face in a scene being imaged, and said right edge is set based on coordinates associated with a right most face in said scene.
1. A computer-implemented method for implementing a zoom function associated with video, the method comprising:
analyzing un-stitched images generated by each of a plurality of imaging sensors for the presence of a face; and
determining a region of interest for the zoom function comprising:
dividing the un-stitched images into a plurality of zones;
setting a left edge for the region of interest to coincide with coordinates associated with a face detected in a left most zone of the un-stitched images;
setting a right edge for the region of interest to coincide with coordinates associated with a face detected in a right most zone of the un-stitched images.
2. The method of claim 1 , further comprising setting scaling parameters based on the region of interest.
3. The method of claim 2 , wherein said scaling parameters includes a scaling factor, and a coordinate associated with each of the faces detected in the left most zone and the right most zone.
4. The method of claim 1 , wherein said analyzing of the un-stitched images comprises implementing a mapping function wherein the zones of said un-stitched images are marked as true in cases a face is detected therein, and false otherwise.
5. The method of claim 4 , wherein each of the plurality of zones is defined by a rectangular zone.
6. The method of claim 1 , wherein said coordinates comprise coordinates associated with a bounding box generated by a face detection engine whenever a face is detected in the un-stitched images.
7. A video system, comprising:
a plurality of imaging sensors;
a plurality of image signal processors, each associated with an imaging sensor;
an image stitching engine to divide unstitched images into a plurality of zones and stitch the unstitched images output by the image signal processors;
a zoom engine configured to determine a region of interest for a zoom function, wherein said region of interest includes a left edge and a right edge, wherein said left edge is set based on coordinates associated with a face detected in a left most zone of the unstitched images, and said right edge is set based on coordinates associated with a right most face detected in a right most zone of the unstitched images.
8. The video system of claim 7 , further comprising a face detection engine for detecting the left most face and the right most face.
9. The video system of claim 8 , further comprising a mapping function implemented in said zoom engine to map the plurality of zones of the unstitched images as true whenever a face is detected therein, and false otherwise.
10. The video system of claim 8 , wherein said the plurality of zones comprises rectangular zones associated with the unstitched images generated by each of the imaging sensor.
11. The video system of claim 8 , wherein the zoom engine is configured to generate scaling parameters for the region of interest.
12. The video system of claim 11 , wherein said scaling parameters comprises a scale factor for the zoom function.
13. The video system of claim 7 , wherein said face detection engine is implemented in each of the image signal processor.