Car detection based on multi-cues integration

We present a novel fast multi-cues based car detection technique in still outdoor images. On the bottom level, two novel area templates based on edge cue and interest points cue are first designed, which can rapidly reject most of the non-car sub-windows at the cost of missing few of the car sub-windows. On the top level, both global structure cue and local texture cue are considered. To character the global structure property the odd Gabor moments are introduced and trained by SVMs. The multi channels even Gabor based local texture property extracted from corner area is modeled as a Gaussian distribution. The final experiment results show that the integration of global structure property and local texture property is more powerful in discrimination between car and non-car objects and a high detection accurate 93% is obtained.