Parallel Approach of Adaptive Image Thresholding Algorithm on GPU

Abstract

Image thresholding is used to segment an image into background and foreground using a given threshold. The threshold can be generated using a specific algorithm instead of a pre-defined value obtained from observation or experiment. However, the algorithm involves per pixel operation, histogram calculation, and iterative procedure to search the optimum threshold that is costly for high-resolution images. In this research, parallel implementations on GPU for three adaptive image thresholding methods, namely Otsu, ISODATA, and minimum cross-entropy, were proposed to optimize their computational times to deal with high-resolution images. The approach involves parallel reduction and parallel prefix sum (scan) techniques to optimize the calculation. The proposed approach was tested on various sizes of grayscale images. The result shows that the parallel implementation of three adaptive image thresholding methods on GPU achieves 4-6 speeds up compared to the CPU implementation, reducing the computational time significantly and effectively dealing with high-resolution images.

Gpu-based and streaming-enabled implementation of pre-processing flow towards enhancing optical character recognition accuracy and efficiency

“…Leptonica supports both Otsu's method [24] and Sauvola's method [25] of thresholding. However, for our implementation, we opted to utilize Otsu's method [24] due to its successful parallelization on GPUs in a previous study [26]. Moreover, Otsu's method exhibits a lower algorithmic complexity as a global thresholding technique compared to Sauvola's local thresholding approach [27].…”

“…Multiple kernel calls are made for each step of the threshold calculation process within Otsu's thresholding kernel. We modify the approach presented in [26] by combining some of the kernel calls. After collecting the initial histogram, we generate zeroorder and first-order cumulative histograms to compute the inter-class variance for each threshold value.…”