An object recognition algorithm using Maximum Margin Correlation Filter and Support Vector Machine

We consider the problem of detecting objects in two dimensional images and propose a new technique which uses Support Vector Machine (SVM) along with Maximum Margin Correlation Filter (MMCF). We have shown that our algorithm detects objects well and is robust with respect to scale changes. Introduction of Support Vector Machine (SVM) helps Maximum Margin Correlation Filter (MMCF) to deal with non-linearly separable data also to some extent. The algorithm also detects same object, if it is found several times at several scales, thus it helps avoiding redundant detection of same object and finally selects the best version of it.