Accurate and Fast Dense Stereo Matching Using Three-Mode Census to Compute Matching Costs and Adaptive Cross Windowing to Aggregate Costs
Two common methods for cost aggregation in local stereo matching are the cross window and the weighted matching window methods. The adaptive weighted window method requires complex calculations to determine the weights, and therefore eliminates the main advantage of local methods, i.e. its high speed. The cross window method is faster than the adaptive weighted window because it is not necessary to define the weights from complex mathematical relationships, but its accuracy is less than the adaptive weighted window method. In the researches that use the cross-matching window, methods such as the absolute value of the color intensity difference or the normal Census transform are used to calculate the cost, the accuracy of which is not as good as the weighted matching window method. In this paper, the three-mode Census transform is used to calculate costs. This method is used along with two other methods, the absolute value of the color intensity difference and the absolute value of the gradient difference, together with the cross aggregation cost, leads to achieving good accuracy also the desired performance speed. Considering the experimental results on the Middlebury standard dataset confirms the desired performance of the proposed algoithm in terms of accuracy and execution speed.
-
Fatigue and drowsiness detection of the car driver based on image processing and artificial intelligence on the mobile phone
Daniyal Haghparast, Alimohammad Fotouhi*
Signal and Data Processing, -
Dense Stereo Matching Based on the Directional Local Binary Pattern
Parisa Bagheri, Ali Fotouhi *
Journal of Electrical Engineering, Winter-Spring 2023