Abstract: Log strata automatic division method as an optimum partitioning one available now is still insufficient to utilize well log patterns and distinguished point information, along with big amount of calculated data and low efficiency, affecting its own reliability. This paper presents a new method for log strata automatic division of optimum partitioning integrated with edge detection. Using this method to find and select log distinguished point and then make an optimum partitioning at the aggregate points could greatly improve the velocity of calculation; meanwhile, the flexibly designing edge detection index means not only to get the maximum interlayer deviation quadratic sum, but also to differentiate the logs style at both sides of dividing point, thus improving the reliability of log strata division. The case of Mesozoic strata division in Jiyang depression shows that this method has a good applied effectiveness.

Key words: log strata division, edge detection, optimum partitioning, Jiyang depression