Mproved Sauvola Threshold for Background Subtraction on Moving Object Detection

Abstract

Image Segmentation is one essential processing on moving object detection. The one of common segmentation methods is thresholding. In this paper, Thresholding method based on adaptive local technique using local mean and standard deviation is known as 'WAN' method. WAN has been inspired by the Sauvola's binarization method and exhibits its robustness and effectiveness when evaluated on low quality document images. The objective of the WAN to enhance the sauvola method and to get a better binarization result and enhance the accuracy. This research aims to produce output value of WAN algorithm. WAN would be compared to other existing adaptive local method like sauvola and niblack. This research is implemented by using matlab and four videos original from camera. The best result calculation error (MSE,PSNR) of WAN method are (0.0011, 53,6655). Overall, the result of WAN method in this paper is more effective and efficient than the other existing method based on MSE and PSNR.

Impact of Wolf Thresholding on Background Subtraction for Human Motion Detection

“…In equation (5) where M is the segmented image of wolf thresholding while N is ground truth image. PSNR in equation (6) symbolize the maximum pixel from the image.…”

“…Segmentation is one of the steps of the background subtraction technique [4]. Thresholding, also known as floating, is the process of converting a grayscale image into a binary image by altering all pixels with a value of zero when below the threshold and one when above the threshold [5]. Thresholding is a simple but effective tool for separating foreground objects from the background, therefore it essential technique for background subtraction [6].…”

Penerapan Image Processing Dalam Sistem Monitoring Ketinggian Air

“…Terdapat banyak teknik dalam proses segmentasi citra salah satunya adalah Thresholding. Thresholding digunakan untuk membagi citra menjadi background dan foreground [13]. Thresholding dilakukan dengan menentukan batas atas dan batas bawah pada nilai suatu warna pada citra, contohnya pada citra dengan color space HSV terdapat 3 batas bawah dan 3 batas atas untuk masing-masing Hue, Saturation, dan Value [14].…”

Modeling a Local Dissimilarity Map With Weibull Distribution–Application to 2-Class and Multi-Class Image Classification

“…For dissimilar image pairs (see Figure (3e)), the LDM appears clearer than for similar image pairs because only a small number of pixels are equal. Images A, B and C are taken from the MNIST data set, initially in grayscale and they have been binarized using the Sauvola threshold [22]. The object is the pixel values equal to 1 and the background is the pixel values equal to 0.…”