This paper points out the obvious wrong segmentation in the existing two-dimensional histogram vertical segmentation method. A new method of two-dimensional histogram oblique segmentation is proposed, in which the histogram is divided into inner, edge and noise parts by four oblique lines paralleled with the main diagonal. The image is segmented according to the sum of point grayscale and neighborhood average grayscale. The method could be used in almost all the two-dimensional histogram threshold algorithms. The formula of the Tsallis–Havrda–Charvát entropy thresholding based on the two-dimensional histogram oblique segmentation and its fast recurring algorithm are deduced. The segmented images and processing time are given. Compared with the original two-dimensional Tsallis–Havrda–Charvát entropy algorithm based on two-dimensional histogram vertical segmentation, the proposed algorithm makes the inner part uniform and the edge accurate in the segmented image, and has better tolerance capability to noise. The processing time is reduced by 5 orders of magnitude.