Out of context at least the small piece of text you cite is very poorly phrased and partially wrong.
First, you should have a look at Solving Hardy Weinberg problems. Take your time and read that post... Done? Good.
More homozygote and less heterozygote individuals in the population
This is wrong as non-assortative mating might be disassortative mating where individuals are attracted to individuals of different genotypes.
However, it is true that population structure will cause excess of homozygotes. This is called the Allee effect. The loss of heterozygosity due to population structure is equal to twice the variance in mean allele frequency among those populations. These details sounds a bit too advance for your needs though so I won't go any further.
Allele frequencies are constant (not in case of negative density dependence)
Genotype frequencies change
Constant over what? Change over what? Not over time necessarily (at least not unless some other assumptions are being made).
I guess what is meant is that if in a given generation mating is not random (due to population structure for example), then you expect the same allele frequency as if it was random but not the same genotype frequency. If it is not clear why, you should probably reread the post "Solving Hardy-Weinberg problems"
Much further discussion in both english and french on @AlissaDeschamps misunderstanding can be found here.