The logistic map can be used to generate pseudo random numbers． However， the data generated directly cannot take on a good uniformity． Based on the characteristic of random numbers generated directly by the logistics map pseudo random generator (LM-PＲNG)，an improved pseudo random number generator (ILM-PＲNG) is designed based on a histogram homogenization method，and the performance of LM-PＲNG and ILM-PＲNG is researched by parameters tests，uniform test and unique test． At 0. 05 level，the results of three kinds of number scales show that the first moment of random numbers generated by LM-PＲNG can pass the tests when N =103，N = 104 and 60% N = 105． The second moment，variance and uniform failed in the test． 93. 3% of the samples can pass the unique test． Meanwhile， random numbers generated by ILM-PＲNG can pass all the tests in this research． The parameter and uniform characters are getting better with the numbers getting larger． As a new kind of PＲNG， the ILM-PＲNG can overcome the lack of the original logistic generator and can be used to generate uniform pseudo random numbers more satisfactorily．