Within n^2, n boundaries were generated from the 1st to the nth, each containing n numbers. Primes less than n^2/2 were multiplied, intersected, and formed composites. At least one prime less than n or in the 1st boundary was used as a factor for the composites between n and n^2, or 2nd and nth boundaries, limiting the number of composites to (2n^2)/λ, where λ represented the wavelength of primes in the 1st boundary. Under these conditions, passively remaining numbers that were not connected to the wave of primes in the 1st boundary were all new primes between the 2nd and nth boundaries. Considering the cause-and-effect relationship among the primes less than n and the composites and new primes between 2nd and nth boundaries, the characteristics of composites could represent the characteristics of primes, and both were defined within a limited n^2 boundary. In this paper, these boundary characteristics were utilized to obtain the average number count per boundary, which led to obtaining the average number of primes per boundary. The average number of primes was multiplied by n boundaries with a coefficient of either β1 or β_√2, denoting the ratio of the number of primes. Using either β1 or β_√2, the number of primes was estimated between 10^6 and 10^28 and compared to the actual number of primes. Considering the relative error between β1 (Average 1.42%: maximum 2.92%, minimum 0.16%) or β_√2 (Ave. 0.37%: max. 0.96, min. 0.04%), it was concluded that the number of primes could be estimated with β_√2, allowing for a relative average error of 0.37%, in an equation of π(n^2)=π(n)∙n/β_√2, where 10^3 ≤ n ≤ 10^14, π(n) was the known number of primes within n, and β_√2 = ln(2√2∙n)/ln(n)+1.