The purpose of this paper is to address the problem of real-time detection of intestinal lumen for the wireless endoscopic capsule and Micro-Ball. With the captured Bayer-format images, a pixel-based local thresholding algorithm is proposed. By threshold processing on local regions of the image, the intestinal lumen can be detected. It has two merits: 1) Automatically acquiring location information of the intestinal lumen; 2) Low memory requirement and low computation complexity suitable for hardware implementation. The performance of the proposed method is verified with three-fold cross validation experiments. It achieves 85.2% accuracy, 78.0% specificity and 91.0% sensitivity, while the memory requirement is reduced to 2% of whole image area and for a N*N image only O(N) operations of add and comparison are needed.