In video surveillance applications, the main aim will be identification of the pedestrians i.e., identifying the position of the persons in each frame of the video. The robust tracking of abrupt motion is a challenging task in computer vision. SVM classifier with HOG features is proposed for effective handling the local-trap problem. A new sampling based scheme is proposed to speed up background separation using SVM classifier. The features were extracted using Co – HOG feature extraction methodology which gives us knowledge about the intensity based informations from the image. The performance of the process is measured. Identification of pedestrians in the video has many applications in surveillance systems. The pedestrians in the video frames were identified by the help of object based detection method. Initially the positions of the objects in the video are identified using background subtraction method. The background subtraction process provides the estimated position location of all the objects in the video. The result of the background subtraction process consists of the object pixels along with the other background pixels in the video. The object pixels alone have to be identified in the obtained pixels. For the identification of the object pixels alone the pixels has to be classified with the help of the extraction of the features and the classification of the pixels into object or others. The extraction of the features can be based on the intensity related information from the pixel or the texture related information. The intensity based method for the feature extraction process is proposed in this process. The Co-HOG method is employed for extracting the intensity relans ted information from the obtained pixels. The obtained features were then classified by employing SVM method. SVM is a supervised learning method that classifies the input data vectors into two categories. In the proposed work the data vectors were classified into object and background. The image pixels that were identified as background were eliminated and the other pixels were obtained. The obtained object pixels were then grouped and the identified positions were then represented in different color inorder to obtain them exactly. Finally the performance of the process is measured by measuring the error rate and accuracy. The measure performance of the process shows that the proposed method is capable for the identification of the object positions in a accurate manner compared to the other existing methods used for the identification of the object positions in the video which is due to the intensity based feature values extracted from the image that denotes the object related information in a better manner.